That's what people said about OpenGL and it looks like that's not going to happen. I'm sure Vulkan will be in turn, replaced by something else. It might take 10-30 years of course.
Not necessarily. OpenGL is a high level abstraction, and as such had a lot of inherent shortcomings. Vulkan is a low level API to use GPUs and as such is far more flexible, and thus it will stay relevant much longer. The only "lower level" you can go is assembly, which will rarely be an attractive option for most developers, since it is not portable and actually increases the effort to support different architectures tremendously.
Think of Vulkan as of C - it is 43 years old now and still very relevant, for exactly that reason - it hits the sweet spot, it is the lowest possible level language with least overhead that is still portable. A big chunk of the codebase today is still C, and the rest is in languages, build mostly with C.
ROFL. Mantle was a waste of VERY valuable resources (that AMD doesn't have much of these days). But you just keep telling yourself mantle has merit. ;) DX12 and glNext were coming anyway, which is why there was no need for Nvidia to do anything (nor amd, but management screwed them again). After the flop that is Win8/8.1, MS had to do something with directX to sell windows 10, which is still win8 with lipstick. Heck they're trying so hard to put lipstick on this pig, it caused them to skip a whole win9...LOL.
But hey, if it makes you feel good, what part of Vulkan code is Mantle? :) I think anandtech is blowing the mantle contribution out of proportion (as an AMD only portal site should I guess, they glorified this waste of money here on anandtech). Since Neil Trevett is running Khronos (and Nvidia's Mobile Ecosystem division), I'm sure NV has some code in there too...LOL.
It really doesn't matter if Mantle was a waste of resources or not. It really wouldn't have mattered if Mantle had no game support. It really wouldn't have mattered if AMD Cancelled it 6 months ago...
The job of Mantle was to light the competitive fires underneath Microsoft's and Kronus's lazy laurels to innovate, they did a fantastic job and now... All consumers in the PC space benefit because of it, regardless if you have an Intel, nVidia or AMD Graphics chip. - That's a mightily impressive "Waste of resources".
Taking nothing away from Mantle. I'm sure Microsoft's strat of unifying windows and Xbox through Direct X was started many years ago and in similar timescales to Valve and AMD.
Mantle was clearly a stop gap. Whether or not it was a waste of resources I don't think is really clear considering Mantle seems to have given at least a minor boost to Vulkan.
I doubt we will see this running on actual hardware till probably end of next year, with games supporting it probably right around two years from today.
Why? Microsoft has no incentive to build in support into Windows. Google's mobile initiatives are so heavily splintered by OEM's that it will probably only be new hardware to actually get the necessary bits in people hands. Apple has always taken their time to support other OpenGL releases. With Metal actually shipping and Apple providing developers with enough tools to make Metal relevant I don't see much benefit for them.
Vulkan could very well be the future of gaming. It will probably take the same amount of time for Electric cars to be more common place.
Mantle was for AMD's sake so they had a good idea of how to start programming next gen APUs/graphics. Just an FYI. It is for all intents and purposes "similar enough".
As I've stated so many times here, tomshardware etc. Mantle is dead. Vulkan will certainly last longer than Mantle did ;) What a waste of money. They spent a ton probably chasing the idea they could help their low quality cpus, isntead of simply making a CPU IPC monster (sans gpu) to dethrone Intel until they could correct and dedicate a ton more to the meager 5% increases we keep getting from Intel's cpu side.
I won't hold my breath on how good NOT SO Freesync will be until reviewed heavily this month when monitors hit. AMD would have shown it running tons of games by now if it worked as good as gsync. But we can pray it's good enough to bring down gsync stuff. I continue to have major hate for AMD's MANAGEMENT. Note no hate for the company, it's workers, or the products - just need better management who cares about CORE products or markets they can actually get a foot into). They need a REAL cpu to bring back some mind share, and more attention to their gpu. APU's won't make them a dime with Intel racing down to stop ARM (armada) from coming up the chain. AMD's apu's just get squeezed out of any pricing power.
They need to quit chasing apu and make a real GPU and CPU killer. Jumping Intel right now with cpu only (as Intel keeps chasing gpu instead of IPC with no pure cpu competition) would allow pricing power as 95% of us gamers just turn off integrated crap and buy a GPU anyway. Again, management is the problem. They have the right engineers to make a great cpu (keller, papermaster etc), but you need to drop the gpu part from it, and dedicate that space to CPU so you can win back gamers (and possibly sell more cards than now to those gamers).
The ATI purchase at 3x what they should have paid was the start of the downfall (no money left for R&D). Then they compounded the problem by wasted resources on Consoles, which make them nothing unless they get to 7yrs selling great where margins improve. But they'll die due to mobile being good enough this xmas at 14nm (X1's replacement etc will be better than xbox360/ps3 already, and worse next year etc as we see 10nm). Then they did it again on Mantle. Mantle was an API doomed from the start and only a last resort to make up for short comings in the cpu.
I don't see a product AMD has on the horizon that will bring pricing power unless they have a cpu monster coming. The gpu side seems too power hungry (among other things, is liquid really required?) and likely just have HBM as "blue crystals" so to speak (marketing garbage that helps nothing now). If we needed more bandwidth today, NV wouldn't have went from 384bit bus to 256 and beat their last cards handily. IT will be needed at some point, or something like it, but not the next gen or two probably. HBM will likely raise costs for AMD jumping too soon as a marketing point and again, makes lower profits if the other guy beats you using standard stuff (and if needed going back to 384 etc). I hope AMD gets bought soon. They have some great people still, they just need to be told by management to do something that makes money.
If all it did was convince Microsoft to build DirectX 12, then it served it's purpose. It's cheaper for AMD to just drop it now and they've achieved their goal.
I agree with some if what you're saying, but not all. Mantle is dead, but at least it started this low level APIA business. I think FreeSync will be quite good (maybe worse than Gsync, but not much), but it another no win situation, imthey don't make any money from it.
Money on consoles is not a waste, it's one of the few places they have steady revenue. The PS4 is based on an app, and it would be bass-ass to see them give it a comercial release (I imagine they are contractually obligated not to). I think APUs is a market they should try to work to their advantage. For casual and older games, APUs are a pretty economically friendly way to game. If AMD can get higher end APUs that are cheap enough, I think they'll have a solid market. Memory bandwidth is an issue, but DDR4 will be standard soon enough.
I think AMD is pretty much stuck as far as CPUs go. For gaming they are ok, but modern i3s are better in some instances, and they almost never beat an i5.
As far as GPUs go, they are still competitive, but only on price. The 280 is a great GPUS that isn't too power hungry and provides a nice 3gb with plenty of bandwidth (compared to the cheap 128 bit, 2gb 960.). But they are too power hungry to be competive for high end, or the low power market (750 ti is a beast that destroys thier low power cards). The only reason to buy an AMD card is price and high res gaming, and only the middle of the road cards (280) make sense in the power segment. I can't imagine AMD is making much money off $180 cards.
Kinda sucks for AMD. They provide some innovation and open standards, and are the only thing keeping INTEL/NVIDIA from becoming a monopoly, but they just can't make money. I think they will just stick to value options like the r9 280 and apu's, and eek out a profit every year.
"Mantle is dead, but at least it started this low level APIA business. " Pfffffftttttttt! Do you think API's just pop into existence in 6 months? More than likely all three have been in development for years before we even heard of Mantle. How could a proprietary API push anyone?
I don't think Metal is going to ever gain any traction, we're looking at an Android dominant market right now. The fact that Vulkan will be available on all popular platforms makes it a clear winner for mobile development, it also makes sense on PC if you want to support Windows, Mac and LInux or any combination of those 3.
Apple's app store still represents significantly more revenue (that is: applications sold) than Android. So the app market is very far from being Android dominated. If you're selling a game, or trying to make any money from a game, then iOS still represents a 60% larger market than Android.
In terms of raw number of applications, only within the last year has Android passed iOS.
It will all boil down to if Apple supports Vulkan. If they do, Metal will fade away. If they don't, it will see heavy use.
Your absolutely right. Metal is available today for iOS devices running Apple's A7 processor or better. That's millions of iPhones (5s, 6, 6+), and iPads (Air, Air 2), that can run Metal apps right now....today. And there are zero (that's right - a big fat goose egg) Android devices that support Vulcan today. Sure, Metal will never gain any traction.
OS X heavily uses OpenGL, and I'm pretty sure Metal is iOS-only. Should be interesting to see if Apple implements Vulkan into OS X in a future release. Probably won't be in 10.11 or even 10.12 though as it's still in development.
Before people flame on AMD and Mantle, read this from the article: In fact Khronos has confirmed that AMD has contributed Mantle towards the development of Vulkan, and though we need to be clear that Vulkan is not Mantle, Mantle was used to bootstrap the process and speed its development, making Vulkan a derivation of sorts of Mantle
TLDR: A small contribution to Vulkan means AMD can cut back on Mantle spending. Win-Win. OpenGL improves, AMD saves money, and DirectX, Metal, etc. get better competition.
Vulcan creates zero competition for Metal API. The only GPGPU driver on iOS is Apple's modified ImgTec GPGPU IP integrated into the A series processors.
Metal is not for OS X.
Vulcan spec will be lucky to be finished by 2015 and won't see full implementation far past 2017.
They've been ``working on this'' since June 2014. In short, they've given themselves 18 months from start to finish on a spec. Meanwhile, OpenGL 4-4.5 is still not completely implemented anywhere, never mind the vendor specific extensions.
I'm all for Vulcan having the potential to stop the ductape world of AMD, Intel, Nvidia extensions and gaming vendors juggling OpenGL/DirectX but Vulcan doesn't change the juggling. When we get an actual commitment from Apple on OS X and all major gaming vendors to support Vulcan, then I'll be impressed.
I wouldn't hold your breath for Apple to support a cross-platform standard. For everyone else, Vulcan looks like a neat idea but will take years to go mainstream.
I don't expect Apple to adopt Vulcan for iOS. I do expect on OS X, but not until OpenGL 4.5 is already past and the gaming industry has flushed out Vulcan on OS X and Windows. Linux will be an afterthought.
I'd like to Phys-X replaced some sort of OpenCL runtime so it could be leveraged on any device. Nvidia has gone through great lengths to lock down its use.
I would throw a GTX 750 in my desktop just to run it with my AMD R290. I bounce back and forth between GPU vendors. Or run physics on the Intel iGPU.
This release is a good step towards making that a possibility.
So if AMD contributed some of its Mantle resources toward Vulkan, then where does that put the future of Mantle? Is Vulkan the future of Mantle, or will AMD continue to develop Mantle on its own?
> to the entire ecosystem of platforms that actively support Khronos’ graphics standards. Which these days is essentially everything outside of the gaming consoles.
Mains concepts of Mantle come from Gallium3D (lighter driver, memory management, and LLVM), LLVMpipe pure sofware driver included in Gallium3D (submodule of Mesa3D open source OpenGL implementation), proven in about 2007, that was this kind of enhancement, pure software GL driver could be more viable than traditionnal architecture with a huge driver. That's why for, SPIR is based on LLVM, SPIR-V is modified to be more adapted to GPU needs, but as said in Khronos slides, a SPIR-V<=>LLVM converter is on the way (will not be too long to be done as SPIR-V is derivated from LLVM).
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
44 Comments
Back to Article
keitaro - Tuesday, March 3, 2015 - link
So will Vulkan be able to live long and prosper?Sorry, had to do it.
damianrobertjones - Tuesday, March 3, 2015 - link
Probably longer than you or I!Flunk - Tuesday, March 3, 2015 - link
That's what people said about OpenGL and it looks like that's not going to happen. I'm sure Vulkan will be in turn, replaced by something else. It might take 10-30 years of course.ddriver - Tuesday, March 3, 2015 - link
Not necessarily. OpenGL is a high level abstraction, and as such had a lot of inherent shortcomings. Vulkan is a low level API to use GPUs and as such is far more flexible, and thus it will stay relevant much longer. The only "lower level" you can go is assembly, which will rarely be an attractive option for most developers, since it is not portable and actually increases the effort to support different architectures tremendously.Think of Vulkan as of C - it is 43 years old now and still very relevant, for exactly that reason - it hits the sweet spot, it is the lowest possible level language with least overhead that is still portable. A big chunk of the codebase today is still C, and the rest is in languages, build mostly with C.
Antronman - Tuesday, March 10, 2015 - link
C isn't used in the development of present-day applications.Derivatives are.
C# was released only 15 years ago, C++ was developed in '79 but only standardized 17 years ago.
dacostafilipe - Tuesday, March 3, 2015 - link
If we then look at some Vulkan code, Vulkan is more Mantle then not!TheJian - Tuesday, March 3, 2015 - link
ROFL. Mantle was a waste of VERY valuable resources (that AMD doesn't have much of these days). But you just keep telling yourself mantle has merit. ;) DX12 and glNext were coming anyway, which is why there was no need for Nvidia to do anything (nor amd, but management screwed them again). After the flop that is Win8/8.1, MS had to do something with directX to sell windows 10, which is still win8 with lipstick. Heck they're trying so hard to put lipstick on this pig, it caused them to skip a whole win9...LOL.But hey, if it makes you feel good, what part of Vulkan code is Mantle? :) I think anandtech is blowing the mantle contribution out of proportion (as an AMD only portal site should I guess, they glorified this waste of money here on anandtech). Since Neil Trevett is running Khronos (and Nvidia's Mobile Ecosystem division), I'm sure NV has some code in there too...LOL.
dacostafilipe - Tuesday, March 3, 2015 - link
You don't believe me? Then what about this :http://forums.anandtech.com/showthread.php?p=37218...
and this
http://forums.anandtech.com/showpost.php?p=3721850...
But hey, I certainly made that up, because, how could somebody like you be wrong ... no, never :P
heffeque - Tuesday, March 3, 2015 - link
That was a very clear "in your face", my good sir.I tip my imaginary hat to you.
bwat47 - Tuesday, March 3, 2015 - link
Yeah, its pretty clear that mantle heavily influenced the low level direction that glnext and dx12 took in the end.Mantle served as an important proof of concept that a low level api is viable on pc. Mantle did its job
piiman - Saturday, March 7, 2015 - link
If only they had started Mantle before dx12 got started. There is zero evidence Mantle influenced dx12 even a little.StevoLincolnite - Thursday, March 5, 2015 - link
It really doesn't matter if Mantle was a waste of resources or not.It really wouldn't have mattered if Mantle had no game support.
It really wouldn't have mattered if AMD Cancelled it 6 months ago...
The job of Mantle was to light the competitive fires underneath Microsoft's and Kronus's lazy laurels to innovate, they did a fantastic job and now... All consumers in the PC space benefit because of it, regardless if you have an Intel, nVidia or AMD Graphics chip. - That's a mightily impressive "Waste of resources".
MojaMonkey - Saturday, March 14, 2015 - link
Taking nothing away from Mantle. I'm sure Microsoft's strat of unifying windows and Xbox through Direct X was started many years ago and in similar timescales to Valve and AMD.Vigilant007 - Thursday, March 5, 2015 - link
Mantle was clearly a stop gap. Whether or not it was a waste of resources I don't think is really clear considering Mantle seems to have given at least a minor boost to Vulkan.I doubt we will see this running on actual hardware till probably end of next year, with games supporting it probably right around two years from today.
Why? Microsoft has no incentive to build in support into Windows. Google's mobile initiatives are so heavily splintered by OEM's that it will probably only be new hardware to actually get the necessary bits in people hands. Apple has always taken their time to support other OpenGL releases. With Metal actually shipping and Apple providing developers with enough tools to make Metal relevant I don't see much benefit for them.
Vulkan could very well be the future of gaming. It will probably take the same amount of time for Electric cars to be more common place.
JonnyDough - Friday, March 6, 2015 - link
Mantle was for AMD's sake so they had a good idea of how to start programming next gen APUs/graphics. Just an FYI. It is for all intents and purposes "similar enough".TheJian - Tuesday, March 3, 2015 - link
http://www.pcper.com/news/Graphics-Cards/GDC-15-AM...As I've stated so many times here, tomshardware etc. Mantle is dead. Vulkan will certainly last longer than Mantle did ;) What a waste of money. They spent a ton probably chasing the idea they could help their low quality cpus, isntead of simply making a CPU IPC monster (sans gpu) to dethrone Intel until they could correct and dedicate a ton more to the meager 5% increases we keep getting from Intel's cpu side.
I won't hold my breath on how good NOT SO Freesync will be until reviewed heavily this month when monitors hit. AMD would have shown it running tons of games by now if it worked as good as gsync. But we can pray it's good enough to bring down gsync stuff. I continue to have major hate for AMD's MANAGEMENT. Note no hate for the company, it's workers, or the products - just need better management who cares about CORE products or markets they can actually get a foot into). They need a REAL cpu to bring back some mind share, and more attention to their gpu. APU's won't make them a dime with Intel racing down to stop ARM (armada) from coming up the chain. AMD's apu's just get squeezed out of any pricing power.
They need to quit chasing apu and make a real GPU and CPU killer. Jumping Intel right now with cpu only (as Intel keeps chasing gpu instead of IPC with no pure cpu competition) would allow pricing power as 95% of us gamers just turn off integrated crap and buy a GPU anyway. Again, management is the problem. They have the right engineers to make a great cpu (keller, papermaster etc), but you need to drop the gpu part from it, and dedicate that space to CPU so you can win back gamers (and possibly sell more cards than now to those gamers).
The ATI purchase at 3x what they should have paid was the start of the downfall (no money left for R&D). Then they compounded the problem by wasted resources on Consoles, which make them nothing unless they get to 7yrs selling great where margins improve. But they'll die due to mobile being good enough this xmas at 14nm (X1's replacement etc will be better than xbox360/ps3 already, and worse next year etc as we see 10nm). Then they did it again on Mantle. Mantle was an API doomed from the start and only a last resort to make up for short comings in the cpu.
I don't see a product AMD has on the horizon that will bring pricing power unless they have a cpu monster coming. The gpu side seems too power hungry (among other things, is liquid really required?) and likely just have HBM as "blue crystals" so to speak (marketing garbage that helps nothing now). If we needed more bandwidth today, NV wouldn't have went from 384bit bus to 256 and beat their last cards handily. IT will be needed at some point, or something like it, but not the next gen or two probably. HBM will likely raise costs for AMD jumping too soon as a marketing point and again, makes lower profits if the other guy beats you using standard stuff (and if needed going back to 384 etc). I hope AMD gets bought soon. They have some great people still, they just need to be told by management to do something that makes money.
tuxfool - Tuesday, March 3, 2015 - link
Did you even read the article? Vulkan is a derivation of mantle.Beany2013 - Tuesday, March 3, 2015 - link
Read the article? That's not the done thing when you have a rambling, foaming at the mouth diatribe to write!Kutark - Tuesday, March 3, 2015 - link
Derivation is probably too strong a word, springboard might be more proper.blaktron - Tuesday, March 3, 2015 - link
Thank goodness you don't run a major electronics firm with actual investors and shareholders.Flunk - Tuesday, March 3, 2015 - link
If all it did was convince Microsoft to build DirectX 12, then it served it's purpose. It's cheaper for AMD to just drop it now and they've achieved their goal.piiman - Saturday, March 7, 2015 - link
Didn't dx12 start development before Mantle?Jumangi - Tuesday, March 3, 2015 - link
Your a graphics tech legend...In your own mind.Creig - Tuesday, March 3, 2015 - link
Somebody needs a bit more fiber their diet.blanarahul - Tuesday, March 3, 2015 - link
*yawn* *rubs eyes* Did I miss something?TallestJon96 - Tuesday, March 3, 2015 - link
I agree with some if what you're saying, but not all. Mantle is dead, but at least it started this low level APIA business. I think FreeSync will be quite good (maybe worse than Gsync, but not much), but it another no win situation, imthey don't make any money from it.Money on consoles is not a waste, it's one of the few places they have steady revenue. The PS4 is based on an app, and it would be bass-ass to see them give it a comercial release (I imagine they are contractually obligated not to). I think APUs is a market they should try to work to their advantage. For casual and older games, APUs are a pretty economically friendly way to game. If AMD can get higher end APUs that are cheap enough, I think they'll have a solid market. Memory bandwidth is an issue, but DDR4 will be standard soon enough.
I think AMD is pretty much stuck as far as CPUs go. For gaming they are ok, but modern i3s are better in some instances, and they almost never beat an i5.
As far as GPUs go, they are still competitive, but only on price. The 280 is a great GPUS that isn't too power hungry and provides a nice 3gb with plenty of bandwidth (compared to the cheap 128 bit, 2gb 960.). But they are too power hungry to be competive for high end, or the low power market (750 ti is a beast that destroys thier low power cards). The only reason to buy an AMD card is price and high res gaming, and only the middle of the road cards (280) make sense in the power segment. I can't imagine AMD is making much money off $180 cards.
Kinda sucks for AMD. They provide some innovation and open standards, and are the only thing keeping INTEL/NVIDIA from becoming a monopoly, but they just can't make money. I think they will just stick to value options like the r9 280 and apu's, and eek out a profit every year.
piiman - Saturday, March 7, 2015 - link
"Mantle is dead, but at least it started this low level APIA business. "Pfffffftttttttt! Do you think API's just pop into existence in 6 months? More than likely all three have been in development for years before we even heard of Mantle. How could a proprietary API push anyone?
TEAMSWITCHER - Tuesday, March 3, 2015 - link
And if Microsoft and Apple favor their own technologies over Vulcan (DX12 and Metal) then I doubt any of this will ever gain traction with developers.Flunk - Tuesday, March 3, 2015 - link
I don't think Metal is going to ever gain any traction, we're looking at an Android dominant market right now. The fact that Vulkan will be available on all popular platforms makes it a clear winner for mobile development, it also makes sense on PC if you want to support Windows, Mac and LInux or any combination of those 3.Guspaz - Tuesday, March 3, 2015 - link
Apple's app store still represents significantly more revenue (that is: applications sold) than Android. So the app market is very far from being Android dominated. If you're selling a game, or trying to make any money from a game, then iOS still represents a 60% larger market than Android.In terms of raw number of applications, only within the last year has Android passed iOS.
It will all boil down to if Apple supports Vulkan. If they do, Metal will fade away. If they don't, it will see heavy use.
TEAMSWITCHER - Tuesday, March 3, 2015 - link
Your absolutely right. Metal is available today for iOS devices running Apple's A7 processor or better. That's millions of iPhones (5s, 6, 6+), and iPads (Air, Air 2), that can run Metal apps right now....today. And there are zero (that's right - a big fat goose egg) Android devices that support Vulcan today. Sure, Metal will never gain any traction.ThreeDee912 - Wednesday, March 4, 2015 - link
OS X heavily uses OpenGL, and I'm pretty sure Metal is iOS-only. Should be interesting to see if Apple implements Vulkan into OS X in a future release. Probably won't be in 10.11 or even 10.12 though as it's still in development.purerice - Tuesday, March 3, 2015 - link
Before people flame on AMD and Mantle, read this from the article:In fact Khronos has confirmed that AMD has contributed Mantle towards the development of Vulkan, and though we need to be clear that Vulkan is not Mantle, Mantle was used to bootstrap the process and speed its development, making Vulkan a derivation of sorts of Mantle
TLDR: A small contribution to Vulkan means AMD can cut back on Mantle spending. Win-Win. OpenGL improves, AMD saves money, and DirectX, Metal, etc. get better competition.
mdriftmeyer - Tuesday, March 3, 2015 - link
Vulcan creates zero competition for Metal API. The only GPGPU driver on iOS is Apple's modified ImgTec GPGPU IP integrated into the A series processors.Metal is not for OS X.
Vulcan spec will be lucky to be finished by 2015 and won't see full implementation far past 2017.
They've been ``working on this'' since June 2014. In short, they've given themselves 18 months from start to finish on a spec. Meanwhile, OpenGL 4-4.5 is still not completely implemented anywhere, never mind the vendor specific extensions.
I'm all for Vulcan having the potential to stop the ductape world of AMD, Intel, Nvidia extensions and gaming vendors juggling OpenGL/DirectX but Vulcan doesn't change the juggling. When we get an actual commitment from Apple on OS X and all major gaming vendors to support Vulcan, then I'll be impressed.
Full disclosure: I own AMD and Apple stock.
stephenbrooks - Tuesday, March 3, 2015 - link
I wouldn't hold your breath for Apple to support a cross-platform standard. For everyone else, Vulcan looks like a neat idea but will take years to go mainstream.mdriftmeyer - Tuesday, March 3, 2015 - link
I don't expect Apple to adopt Vulcan for iOS. I do expect on OS X, but not until OpenGL 4.5 is already past and the gaming industry has flushed out Vulcan on OS X and Windows. Linux will be an afterthought.Antronman - Tuesday, March 10, 2015 - link
Well if it means anything, Source 2 is going to have Vulkan support right out of the box.People expect it will be released around The International.
eanazag - Tuesday, March 3, 2015 - link
I'd like to Phys-X replaced some sort of OpenCL runtime so it could be leveraged on any device. Nvidia has gone through great lengths to lock down its use.I would throw a GTX 750 in my desktop just to run it with my AMD R290. I bounce back and forth between GPU vendors. Or run physics on the Intel iGPU.
This release is a good step towards making that a possibility.
mdriftmeyer - Tuesday, March 3, 2015 - link
Bullet 3 is an OpenCL implementation Physics library:https://github.com/bulletphysics/bullet3
SirKnobsworth - Tuesday, March 3, 2015 - link
So if AMD contributed some of its Mantle resources toward Vulkan, then where does that put the future of Mantle? Is Vulkan the future of Mantle, or will AMD continue to develop Mantle on its own?Krysto - Thursday, March 5, 2015 - link
> to the entire ecosystem of platforms that actively support Khronos’ graphics standards. Which these days is essentially everything outside of the gaming consoles.PS4 supports OpenGL, no?
loguerto - Friday, March 6, 2015 - link
This is what mantle was intended to become.JonnyDough - Friday, March 6, 2015 - link
All you had to do was say "open" and it has my support. Screw all these companies and their proprietary crap.Popolon - Thursday, August 20, 2015 - link
Mains concepts of Mantle come from Gallium3D (lighter driver, memory management, and LLVM), LLVMpipe pure sofware driver included in Gallium3D (submodule of Mesa3D open source OpenGL implementation), proven in about 2007, that was this kind of enhancement, pure software GL driver could be more viable than traditionnal architecture with a huge driver. That's why for, SPIR is based on LLVM, SPIR-V is modified to be more adapted to GPU needs, but as said in Khronos slides, a SPIR-V<=>LLVM converter is on the way (will not be too long to be done as SPIR-V is derivated from LLVM).