I have a bad feeling about Mantle. It's had a pretty lackluster outing, and DirectX 12 is targeting it directly. Even if it doesn't reach Mantle levels of efficiency, at worst it'll be almost-Mantle-level plus it'll BE DirectX. This really won't end well.
More like Mantle was rushed to market to beat DX12 and OpenGL that have already planned these features. NVIDIA has been able to beat Mantle using similar features that are already in DX11.
This is the most completely false statement I have ever read. Mantle is offers a massive increase in minimum framerate as well as frame delivery smoothness that DX11 can absolutely not achieve on AMD or Nvidia cards. Anyone who has played BF4 with Mantle can attest to that. In my testing on the worst part of Siege of Shanghai I get 120fps (capped with r9 290) with Mantle, and 58fps on a gtx780, and 59 fps with the r9 290 in dx11. This is entirely due to the MASSIVE cpu bottleneck in DX11.
I just read that link, Wreckage, and it doesn't really support your statements. That looks an awful lot like a combination of broad driver improvements plus game-specific optimizations. Both teams already do that sort of work on a regular basis. Both DX and Mantle performance can improve over time with new drivers. Also, this is in that same article:
"To its credit, Nvidia did clarify that its new driver has a higher average frame rate, but still has more slow frames than Mantle"
This is what Vulgar was talking about. Also keep in mind Nvidia wasn't using a low-end or older CPU. Mantle isn't perfect but it is pushing MS to release a low-level DX in the next year or two. In the meantime, games utilizing Mantle can get substantial boosts in framerate, especially if you have a slower CPU. I'm all for anything that will increase PC gaming userbase, and being able to play games with cheaper hardware helps enable that. Speaking of which, Mantle is very helpful for GCN-equipped APUs.
He's not going to acknowledge anything you say. It doesn't fit in with his agenda, which is to downplay and disparage AMD vs. nVidia at all opportunities. Accuracy of his statements or the facts are irrelevant.
Really then why did NV have to come up with their new OGL extensions? To get Mantle like performance in OGL. AMD has already added the new OGL extensions to their drivers.
You're completely wrong there. Nvidia's DX11 optimizations work with the Shader Cache, not with a draw call reduction, which is what Mantle does and what DX12 will also do.
You are right, of course...;) AMD is relishing setting a fire under Microsoft's rear in order to nudge the company back on track with D3d development. Had Microsoft not been so lackluster in its continuing support of D3d (since D3d 10, really) AMD would not have had to do this at all. I am sure that given a choice, AMD would rather not spend its own money to step into Microsoft's position, here--but they felt they had little choice, obviously, except to go ahead with Mantle. Comments that AMD was trying to "beat Microsoft to the punch" make no sense at all, because Mantle is not a D3d replacement (it only supports AMD gpus.) Mantle has nothing to do whatever with D3d development--which is entirely up to Microsoft.
This is an extremely short-sighted way to look at this. And obviously Mantle is a D3D replacement lol, what are you even on about?
Let's please keep in mind that DX11 is almost five years old. This is not a scenario where AMD woke a sleeping DirectX Dragon, this is a scenario where Microsoft has already been laser-focused on legitimizing their place in the mobile space with Windows 8 and Phone 8, and now DX12. This was happening whether AMD did a single thing or not.
I really wouldn't be surprised if Mantle wasn't actually based on DX12 in at least some loose or conceptual manner. That honestly makes more sense than the idea that AMD designed its own graphics API for its own hardware and ended up with the results that we're seeing today. The results are more indicative of a design that's more generalized as opposed to a design that's proprietary.
So far, it's virtually Mantle's sole effect that graphics become less CPU-bound. DirectX would obviously do that. Obviously. But AMD's own software on AMD's own hardware? Wouldn't Mantle take more advantage of that sort of integration? Either AMD is really stupid, or the design of Mantle echos designs seen elsewhere, either in OpenGL developments or in DirectX developments, probably the latter as I'm guessing some kind of DX beta code for mobile/Xbox One was floating around somewhere.
Microsoft has been working on DX12 for over 4 years. The latest OpenGL spec was announced prior to the Mantle announcement. AMD simply copied what was coming and rushed a beta on the market
Nah, based on comments from people involved with Mantle, DX12 is almost certainly spawned from Mantle, there's no way AMD could have had something ready faster than MS could. Most likely MS has been involved with Mantle when AMD started work on it, and DX12 will simply be Mantle + MS & NV extensions.
AMD could easy done it faster since they ONLY have to focus on making it work with THEIR gpu's. MS on the other hand has to make it work accross many diff gpu's and setup's and can't limit it to just 1 thing. It was sad when AMD CLAIMED MS had nothing in the pipe for directX but that was clearly a lie by AMD to sell gpu's as they no way they didn't know MS was working on DX12 if its been in the works for 4 years.
WOW Someone should have informed AMD and Nvidia that MS was working on DX12 this whole time. Meanwhile NV revamped OGL and is pushing it forward into game engines at a breakneck pace AMD matched NV in OGL 4.4 support. AMD developed Mantle and it is being adopted very quickly. AMD commits to fixing their Linux drivers.
Yup looks like DX 12 development which would have closely involved both AMD and NV has been in the works this whole time........please That whole 4 years is bullshit.
Still, that would be a win for AMD as they have problems with (relatively big) overhead in their drivers. Reducing the overhead either by the use of Mantle or DirectX 12 reduces the gap between Radeons and GeForces in CPU limited games.
If DirectX 12 will work as good as Mantle on Radeons then it will be even bigger win than Mantle adoption as AMD wouldn't have to promote and support (and thus spend money on) another proprietary standard.
For a company that's already in such rough battles, it just seems like a waste to spend so much R&D, marketing, and face on something that's just going to be done better by someone else a year later.
It would be too late by then. OGL is the future of cross platform compatibility. Why would developers hamstring themselves with DX12 when they can use OGL and have every OS supported.
This many developers jumping on board can only mean AMD plans on bringing Mantle to Linux. Otherwise they would only focus on OGL 4.4.
Just using OGL is not enough, because system capabilities vary so very much. You can't just port games between phone, tablet, PC, console. Different OSes different hardware. Everyone has a rendering abstraction these days, so they don't care what API they use.
DX12 has been in the works for ages alongside WDDM 2.0. A decent chunk of performance gains come from outside of the graphics API itself. There has been very little need for DX12 anyway. Mantle is hot air. Devs target lowest common denominator, which is 0% of commentators on this website.
Mantle and DX12 are barely related. OGL is barely used.
What's of greater interest is OpenGL doing much of the same thing thanks to some clever changes in calling OpenGL (see the Nvidia/Intel slides from GDC 2014 if you want more info on that), and with OpenGL making a return (Steam on Linux, Android, iOS, OSX), it may be an even bigger disruption than DX12 for AMD's Mantle plans.
Why? Competition is making MS update DX. Mantle and OpenGL. Glad to see something creating changes, not sure if Mantle, DX or OpenGL actually started development first but good to see them!
OpenGL was first, it has been done for a few years. Mantle just caught up in Beta, DX12 xmas 2015 probably. OpenGL (NOT in beta) can be used today as my link shows in detail. You don't have to wait for it or beta test it. It's DONE. Usable today. GDC 2014 was an eye opener for many I guess, but they could have been making use of faster draw calls for years. As the guy points out, they just didn't know how. Now that they do what's the point? No need to wait for DX12 or use Mantle right?
If OpenGL is done and can already do this for years, Mantle came 2nd right? And in beta as of today. DX12 was on the drawing board ages ago too. I don't think it had anything to do with Mantle's development pushing them, maybe just made them unveil it a bit sooner. Since they'll tie it to Win8/9 for sure anyway, my guess is it ends up like DX10, basically skipped as OpenGL starts to entrench for the next 2yrs with Valve/Google/Apple etc pushing it for ALL their OS systems.
1) Safe multithreading? 2) Access to Command Queues? 3) Access to individual engines on GPU? 4) Access to indivudual GPUs individual engines? (UberMultiGPU that require NO intervention from GPU vendor). 5) Pre generation of shaders, assets, textures, whatever. (However that is also implementation problem, and thus can be solved to some great extend without new specs). 6) Unified memory both for content and for CPU/GPU memory. (This is hard to solve as not all GPU vendors have solutions here).
AMD put a lot of work into APU's, hybrid computing and such stuff. Not DX nor OGL match 100% here. Both need extension, and as one can guess, other vendors wont allow AMD to set standard, nor they will go for DX/OGL features they can not support.
Mantle is a good thing. Without it DirectX 12 wouldn't even be trying to address these issues. What we have seen thus far is that Mantle really helps slower CPU systems. It basically makes an AMD CPU and GPU combo competitive.
I used to think DirectX 12 would kill Mantle and hurt AMD but then I realized what a genius move Mantle really was. If Mantle catches on that's great AMD sells more graphics cards AND CPUs, if DirectX 12 kills Mantle by also eliminating the CPU bottle neck that's great too because AMD processors just became a viable alternative for hardcore gamers.
if Mantle adoption grows, for AMD will be a big win, for both CPU and GPU if the DX12 brings the same improvement of Mantle, for AMD (cpu) will be a very big win also, because the more powerfull cpu like intel i7 series will be worthless and the AMD APU will be the best choices for gaming (light to medium and why not, some probably also hardcore with some future iteration of AMD APU)
OpenGL does the same stuff as they showed already. DX12 may not do it for a while (late 2015?) but OpenGL does it already and has for a few years. So devs, write for AMD's small subset of their own cards (only GCN) or just use OpenGL which runs on everything easily portable to everywhere. I really thought AMD would have dropped this by now and started spending more of this money on DRIVERS for all their current cards! You know, to avoid a phase4 soon etc or more driver phases for the next chips... :(
http://blogs.nvidia.com/blog/2014/03/20/opengl-gdc... Watch the vid at bottom AMD ;) One of your people was involved in this show, so why keep pushing proprietary for something that is already working on all cards TODAY? Proprietary crap works when you get there first and are special (doing something others can't, thus a NEED for your tech, like Cuda, it was their only choice back then, so it got entrenched). It doesn't work when there is an established product already out that does everything your proprietary stuff does cheaper. This would make more sense if OpenGL didn't exist, and only options were Mantle or DX12 which is ages away (thus Mantle would have a chance to get entrenched). SteamOS will be pushing OpenGL, and same for Android etc if a game needs a high number of draw calls etc.
OpenGL is not the same thing as Mantle. Mantle is a low level API that operates through a much thinner driver layer and allows the devs direct control over GPU functions.
OGL does not multithread the GPU. It reduces CPU overhead a shitload. And now that NV made those extensions for OGL AMD can use those techniques in Mantle.
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.
51 Comments
Back to Article
Ortanon - Thursday, May 1, 2014 - link
I have a bad feeling about Mantle. It's had a pretty lackluster outing, and DirectX 12 is targeting it directly. Even if it doesn't reach Mantle levels of efficiency, at worst it'll be almost-Mantle-level plus it'll BE DirectX. This really won't end well.Goty - Thursday, May 1, 2014 - link
Won't it? Who's to say that AMD's Mantle push isn't at least in part responsible for some of the feature targets of DX12?Wreckage - Thursday, May 1, 2014 - link
More like Mantle was rushed to market to beat DX12 and OpenGL that have already planned these features. NVIDIA has been able to beat Mantle using similar features that are already in DX11.VulgarDisplay - Thursday, May 1, 2014 - link
This is the most completely false statement I have ever read. Mantle is offers a massive increase in minimum framerate as well as frame delivery smoothness that DX11 can absolutely not achieve on AMD or Nvidia cards. Anyone who has played BF4 with Mantle can attest to that. In my testing on the worst part of Siege of Shanghai I get 120fps (capped with r9 290) with Mantle, and 58fps on a gtx780, and 59 fps with the r9 290 in dx11. This is entirely due to the MASSIVE cpu bottleneck in DX11.Wreckage - Thursday, May 1, 2014 - link
NVIDIA's latest driver makes use of code already in DX11 to reduce overhead.tomshardware.com/news/nvidia-driver-update-direct3d-optimization,26381.html
No need for a separate proprietary vendor locked API
Alexvrb - Thursday, May 1, 2014 - link
I just read that link, Wreckage, and it doesn't really support your statements. That looks an awful lot like a combination of broad driver improvements plus game-specific optimizations. Both teams already do that sort of work on a regular basis. Both DX and Mantle performance can improve over time with new drivers. Also, this is in that same article:"To its credit, Nvidia did clarify that its new driver has a higher average frame rate, but still has more slow frames than Mantle"
This is what Vulgar was talking about. Also keep in mind Nvidia wasn't using a low-end or older CPU. Mantle isn't perfect but it is pushing MS to release a low-level DX in the next year or two. In the meantime, games utilizing Mantle can get substantial boosts in framerate, especially if you have a slower CPU. I'm all for anything that will increase PC gaming userbase, and being able to play games with cheaper hardware helps enable that. Speaking of which, Mantle is very helpful for GCN-equipped APUs.
3DVagabond - Thursday, May 1, 2014 - link
He's not going to acknowledge anything you say. It doesn't fit in with his agenda, which is to downplay and disparage AMD vs. nVidia at all opportunities. Accuracy of his statements or the facts are irrelevant.Alexey291 - Saturday, May 3, 2014 - link
He does have 1 point though. There is NOTHING good about a vendor locked API. But its not like NV doesn't have any of those :)The rest - yeah he seems to have a very clear agenda in mind.
at80eighty - Thursday, May 1, 2014 - link
it truly is a shame youre allowed to still post on this site since your VC&G threadshitting daysGizmosis350k - Friday, May 2, 2014 - link
That is a rude thing to say sirTheJian - Thursday, May 1, 2014 - link
OpenGL already has had this for a few years. People just didn't know it was in there or how to use it.grndzro7 - Friday, May 2, 2014 - link
Really then why did NV have to come up with their new OGL extensions?To get Mantle like performance in OGL. AMD has already added the new OGL extensions to their drivers.
Gizmosis350k - Friday, May 2, 2014 - link
OpenGL was at first INCREDIBLY slow and unwieldy so, I don't blame 'emformulav8 - Friday, May 2, 2014 - link
If it was the other way around you would be saying how lovely NVidia and Mantle is.Vayra - Thursday, June 5, 2014 - link
You're completely wrong there. Nvidia's DX11 optimizations work with the Shader Cache, not with a draw call reduction, which is what Mantle does and what DX12 will also do.Nice troll though.
WaltC - Thursday, May 1, 2014 - link
You are right, of course...;) AMD is relishing setting a fire under Microsoft's rear in order to nudge the company back on track with D3d development. Had Microsoft not been so lackluster in its continuing support of D3d (since D3d 10, really) AMD would not have had to do this at all. I am sure that given a choice, AMD would rather not spend its own money to step into Microsoft's position, here--but they felt they had little choice, obviously, except to go ahead with Mantle. Comments that AMD was trying to "beat Microsoft to the punch" make no sense at all, because Mantle is not a D3d replacement (it only supports AMD gpus.) Mantle has nothing to do whatever with D3d development--which is entirely up to Microsoft.Ortanon - Thursday, May 1, 2014 - link
This is an extremely short-sighted way to look at this. And obviously Mantle is a D3D replacement lol, what are you even on about?Let's please keep in mind that DX11 is almost five years old. This is not a scenario where AMD woke a sleeping DirectX Dragon, this is a scenario where Microsoft has already been laser-focused on legitimizing their place in the mobile space with Windows 8 and Phone 8, and now DX12. This was happening whether AMD did a single thing or not.
I really wouldn't be surprised if Mantle wasn't actually based on DX12 in at least some loose or conceptual manner. That honestly makes more sense than the idea that AMD designed its own graphics API for its own hardware and ended up with the results that we're seeing today. The results are more indicative of a design that's more generalized as opposed to a design that's proprietary.
So far, it's virtually Mantle's sole effect that graphics become less CPU-bound. DirectX would obviously do that. Obviously. But AMD's own software on AMD's own hardware? Wouldn't Mantle take more advantage of that sort of integration? Either AMD is really stupid, or the design of Mantle echos designs seen elsewhere, either in OpenGL developments or in DirectX developments, probably the latter as I'm guessing some kind of DX beta code for mobile/Xbox One was floating around somewhere.
Wreckage - Thursday, May 1, 2014 - link
Microsoft has been working on DX12 for over 4 years. The latest OpenGL spec was announced prior to the Mantle announcement. AMD simply copied what was coming and rushed a beta on the marketMedallish - Thursday, May 1, 2014 - link
Nah, based on comments from people involved with Mantle, DX12 is almost certainly spawned from Mantle, there's no way AMD could have had something ready faster than MS could. Most likely MS has been involved with Mantle when AMD started work on it, and DX12 will simply be Mantle + MS & NV extensions.B3an - Thursday, May 1, 2014 - link
You're talking out of your ass.arbit3r - Tuesday, May 6, 2014 - link
AMD could easy done it faster since they ONLY have to focus on making it work with THEIR gpu's. MS on the other hand has to make it work accross many diff gpu's and setup's and can't limit it to just 1 thing. It was sad when AMD CLAIMED MS had nothing in the pipe for directX but that was clearly a lie by AMD to sell gpu's as they no way they didn't know MS was working on DX12 if its been in the works for 4 years.grndzro7 - Friday, May 2, 2014 - link
WOW Someone should have informed AMD and Nvidia that MS was working on DX12 this whole time.Meanwhile NV revamped OGL and is pushing it forward into game engines at a breakneck pace
AMD matched NV in OGL 4.4 support.
AMD developed Mantle and it is being adopted very quickly.
AMD commits to fixing their Linux drivers.
Yup looks like DX 12 development which would have closely involved both AMD and NV has been in the works this whole time........please That whole 4 years is bullshit.
formulav8 - Friday, May 2, 2014 - link
He hates anything that could show NVidia in a bad light and AMD in a good light. I have no idea how people can worship some company.Alexey291 - Saturday, May 3, 2014 - link
I'm surprised they didn't say 6 years seeing how it took them this long to pull their finger out since last version (iirc 11 came out in 2008?)Wibowit - Thursday, May 1, 2014 - link
Still, that would be a win for AMD as they have problems with (relatively big) overhead in their drivers. Reducing the overhead either by the use of Mantle or DirectX 12 reduces the gap between Radeons and GeForces in CPU limited games.If DirectX 12 will work as good as Mantle on Radeons then it will be even bigger win than Mantle adoption as AMD wouldn't have to promote and support (and thus spend money on) another proprietary standard.
Ortanon - Thursday, May 1, 2014 - link
For a company that's already in such rough battles, it just seems like a waste to spend so much R&D, marketing, and face on something that's just going to be done better by someone else a year later.pidgin - Thursday, May 1, 2014 - link
nopiroroadkill - Thursday, May 1, 2014 - link
A company that's supplying chips to the Wii U (okay, that's a joke), the Xbox One (sorry, another joke), and the PS4 (OK I suppose).WithoutWeakness - Thursday, May 1, 2014 - link
So bravegrndzro7 - Friday, May 2, 2014 - link
It would be too late by then. OGL is the future of cross platform compatibility.Why would developers hamstring themselves with DX12 when they can use OGL and have every OS supported.
This many developers jumping on board can only mean AMD plans on bringing Mantle to Linux. Otherwise they would only focus on OGL 4.4.
djgandy - Friday, May 2, 2014 - link
Just using OGL is not enough, because system capabilities vary so very much. You can't just port games between phone, tablet, PC, console. Different OSes different hardware. Everyone has a rendering abstraction these days, so they don't care what API they use.djgandy - Friday, May 2, 2014 - link
DX12 has been in the works for ages alongside WDDM 2.0. A decent chunk of performance gains come from outside of the graphics API itself. There has been very little need for DX12 anyway. Mantle is hot air. Devs target lowest common denominator, which is 0% of commentators on this website.Mantle and DX12 are barely related. OGL is barely used.
ZeDestructor - Thursday, May 1, 2014 - link
What's of greater interest is OpenGL doing much of the same thing thanks to some clever changes in calling OpenGL (see the Nvidia/Intel slides from GDC 2014 if you want more info on that), and with OpenGL making a return (Steam on Linux, Android, iOS, OSX), it may be an even bigger disruption than DX12 for AMD's Mantle plans.grndzro7 - Friday, May 2, 2014 - link
Except AMD is on board with OGL 4.4It would not affect AMD at all.
JackNSally - Thursday, May 1, 2014 - link
Why? Competition is making MS update DX. Mantle and OpenGL. Glad to see something creating changes, not sure if Mantle, DX or OpenGL actually started development first but good to see them!TheJian - Thursday, May 1, 2014 - link
OpenGL was first, it has been done for a few years. Mantle just caught up in Beta, DX12 xmas 2015 probably. OpenGL (NOT in beta) can be used today as my link shows in detail. You don't have to wait for it or beta test it. It's DONE. Usable today. GDC 2014 was an eye opener for many I guess, but they could have been making use of faster draw calls for years. As the guy points out, they just didn't know how. Now that they do what's the point? No need to wait for DX12 or use Mantle right?If OpenGL is done and can already do this for years, Mantle came 2nd right? And in beta as of today. DX12 was on the drawing board ages ago too. I don't think it had anything to do with Mantle's development pushing them, maybe just made them unveil it a bit sooner. Since they'll tie it to Win8/9 for sure anyway, my guess is it ends up like DX10, basically skipped as OpenGL starts to entrench for the next 2yrs with Valve/Google/Apple etc pushing it for ALL their OS systems.
grndzro7 - Friday, May 2, 2014 - link
OGL couldn't do it. NV had to make new OGL extensions.djgandy - Friday, May 2, 2014 - link
Yeah because if Microsoft didn't react with DX12 everyone would uninstall Windows and go with AMDOS?przemo_li - Monday, May 5, 2014 - link
OpenGL do not have many answers yet.1) Safe multithreading?
2) Access to Command Queues?
3) Access to individual engines on GPU?
4) Access to indivudual GPUs individual engines? (UberMultiGPU that require NO intervention from GPU vendor).
5) Pre generation of shaders, assets, textures, whatever. (However that is also implementation problem, and thus can be solved to some great extend without new specs).
6) Unified memory both for content and for CPU/GPU memory. (This is hard to solve as not all GPU vendors have solutions here).
AMD put a lot of work into APU's, hybrid computing and such stuff.
Not DX nor OGL match 100% here. Both need extension, and as one can guess, other vendors wont allow AMD to set standard, nor they will go for DX/OGL features they can not support.
przemo_li - Monday, May 5, 2014 - link
PSI'm for OpenGL.
But I think that Mantle can be placed on solid 3rd behind Mesa, Valve involvement, when counting beneficial influence on OpenGL.
eanazag - Thursday, May 1, 2014 - link
Mantle is a good thing. Without it DirectX 12 wouldn't even be trying to address these issues. What we have seen thus far is that Mantle really helps slower CPU systems. It basically makes an AMD CPU and GPU combo competitive.chris200x9 - Thursday, May 1, 2014 - link
I used to think DirectX 12 would kill Mantle and hurt AMD but then I realized what a genius move Mantle really was. If Mantle catches on that's great AMD sells more graphics cards AND CPUs, if DirectX 12 kills Mantle by also eliminating the CPU bottle neck that's great too because AMD processors just became a viable alternative for hardcore gamers.przemo_li - Monday, May 5, 2014 - link
And DX is 2y from now.That is a lot of time for making sure Game Devs know Your capabilities. Like Your ways, and curse competition when it lack X or Y feature ;)
pidgin - Thursday, May 1, 2014 - link
dying to see some HSA goodness in gaming scene, using iGPUs to take load off from dGPU etc etcextide - Thursday, May 1, 2014 - link
I'd love to see APU + dGPU, and use the iGPU for extra compute on top of the CPU cores (physics, AI, etc), and the dGPU for graphics.Dirty_Punk - Thursday, May 1, 2014 - link
if Mantle adoption grows, for AMD will be a big win, for both CPU and GPUif the DX12 brings the same improvement of Mantle, for AMD (cpu) will be a very big win also, because the more powerfull cpu like intel i7 series will be worthless and the AMD APU will be the best choices for gaming (light to medium and why not, some probably also hardcore with some future iteration of AMD APU)
TheJian - Thursday, May 1, 2014 - link
OpenGL does the same stuff as they showed already. DX12 may not do it for a while (late 2015?) but OpenGL does it already and has for a few years. So devs, write for AMD's small subset of their own cards (only GCN) or just use OpenGL which runs on everything easily portable to everywhere. I really thought AMD would have dropped this by now and started spending more of this money on DRIVERS for all their current cards! You know, to avoid a phase4 soon etc or more driver phases for the next chips... :(http://blogs.nvidia.com/blog/2014/03/20/opengl-gdc...
Watch the vid at bottom AMD ;) One of your people was involved in this show, so why keep pushing proprietary for something that is already working on all cards TODAY? Proprietary crap works when you get there first and are special (doing something others can't, thus a NEED for your tech, like Cuda, it was their only choice back then, so it got entrenched). It doesn't work when there is an established product already out that does everything your proprietary stuff does cheaper. This would make more sense if OpenGL didn't exist, and only options were Mantle or DX12 which is ages away (thus Mantle would have a chance to get entrenched). SteamOS will be pushing OpenGL, and same for Android etc if a game needs a high number of draw calls etc.
3DVagabond - Friday, May 2, 2014 - link
OpenGL is not the same thing as Mantle. Mantle is a low level API that operates through a much thinner driver layer and allows the devs direct control over GPU functions.grndzro7 - Friday, May 2, 2014 - link
OGL does not multithread the GPU. It reduces CPU overhead a shitload.And now that NV made those extensions for OGL AMD can use those techniques in Mantle.
diogodx - Friday, May 2, 2014 - link
More fanboys spreading FUD. The OpenGL CPU overhead redutcion extention was created by AMD.The extension "ARB_multi_draw_indirect" that Nvidia show in their market was created by Graham Selles from AMD.
http://www.opengl.org/registry/specs/ARB/multi_dra...
przemo_li - Monday, May 5, 2014 - link
Based on feedback from G-Truck. And that was created in mere weeks.Though with Mantle point made is that NO SUCH INTERVENTION from AMD/Nvidia/Intel would be needed.
That is biggest selling point of Mantle lower level approach.
Much less dependency on GPU vendor cooperating.