AMD are working very closely with the Khronos Group for the next generation OpenGL API, and have even given them access to see what makes their own Mantle API tick.
Techreport managed to speak to Richard Huddy from AMD about it all:
That's pretty good of AMD to offer it up like that, but to what end? What would be the point of having an OpenGL API that very closely mimics the Mantle API? You would then have two competing API's that are near the same.
It's also interesting to note that AMD's Mantle helped shape the new DirectX:
Looks like the graphics API space is going to get more and more interesting as time goes on.
It still remains to be seen if Mantle will come to Linux or not, but personally I am still unsure at just how far Mantle will go. AMD are claiming that a lot of developers are going to support it, so it could be around for quite some time.
What do you think to this?
Techreport managed to speak to Richard Huddy from AMD about it all:
QuoteHuddy told us AMD has done a "great deal of work" with the Khronos Group, the stewards of the OpenGL spec, on OpenGL Next. AMD has given the organization unfettered access to Mantle and told them, in so many words, "This is how we do it. If you want to take the same approach, go ahead." Khronos is free to take as many pages as it wants out of the Mantle playbook, and AMD will impose no restrictions, nor will it charge any licensing fees.
That's pretty good of AMD to offer it up like that, but to what end? What would be the point of having an OpenGL API that very closely mimics the Mantle API? You would then have two competing API's that are near the same.
It's also interesting to note that AMD's Mantle helped shape the new DirectX:
QuoteWhile Huddy didn't say how closely OpenGL Next might mirror Mantle, he repeated the contention that Mantle shaped DirectX 12's development. We expressed some doubts about that contention when we addressed it earlier this year, but Huddy was adamant. Development on DirectX 12's new features may have begun before Mantle, he said, but the "real impetus" for DX12's high-throughput layer came from the AMD API.
Looks like the graphics API space is going to get more and more interesting as time goes on.
It still remains to be seen if Mantle will come to Linux or not, but personally I am still unsure at just how far Mantle will go. AMD are claiming that a lot of developers are going to support it, so it could be around for quite some time.
What do you think to this?
Some you may have missed, popular articles from the last month:
9 comments
Well, if OpenGL 5 comes to resemble Mantle (which resembles OpenGL anyway), technically speaking the assertion "a lot of developers are going to support it" would be true.
0 Likes
If OpenGL Next would closely resemble Mantle, AMD would have a headstart in implementing it.
On the other hand, if it's quite close, it might make Mantle obsolete.
On the other hand, if it's quite close, it might make Mantle obsolete.
0 Likes
Personally, I feel extremely excited with GL-Next. The direction that Khronos follows is to create a spec where the application developers have the power and the control over the GPUs. The driver layer is going to be thin and way less complicated than the current GL.
Some notes on the GL-Next initiative:
- AMD made a major contribution by giving Mantle to Khronos (I believe DICE contributed to Mantle as well).
- There is a consensus on the major goals of GL-Next between Khronos members. This gives a great momentum on the whole effort.
- It's really nice to see game developers contributing with proposals and feedback. Valve, Epic, Unity, DICE, Blizzard etc. Valve is sponsoring some stuff as well!
So next time, carefully consider to whom you give your money to ;)
Some notes on Mantle:
- Obviously, Mantle is close to AMD HW and it will require some changes to accommodate other HW.
- It not going to be easy to code with Mantle. Some "nice" driver abstractions will become an application problem. And GL 4.5 has tons of those abstractions.
- Drivers will have far better quality. Implementing something like Mantle is way easier that implementing the whole GL 4.5 stack.
- The opensource drivers will probably be up to speed a lot quicker than before.
Some notes on the GL-Next initiative:
- AMD made a major contribution by giving Mantle to Khronos (I believe DICE contributed to Mantle as well).
- There is a consensus on the major goals of GL-Next between Khronos members. This gives a great momentum on the whole effort.
- It's really nice to see game developers contributing with proposals and feedback. Valve, Epic, Unity, DICE, Blizzard etc. Valve is sponsoring some stuff as well!
So next time, carefully consider to whom you give your money to ;)
Some notes on Mantle:
- Obviously, Mantle is close to AMD HW and it will require some changes to accommodate other HW.
- It not going to be easy to code with Mantle. Some "nice" driver abstractions will become an application problem. And GL 4.5 has tons of those abstractions.
- Drivers will have far better quality. Implementing something like Mantle is way easier that implementing the whole GL 4.5 stack.
- The opensource drivers will probably be up to speed a lot quicker than before.
0 Likes
QuoteThat's pretty good of AMD to offer it up like that, but to what end? What would be the point of having an OpenGL API that very closely mimics the Mantle API? You would then have two competing API's that are near the same.
My first thought is that AMD are fighting a losing battle on both the CPU and GPU front, being quite clearly second best at both. "Helping" both OpenGL and DirectX by giving them code and direction tailored to AMD hardware will bridge that gap, and might even push them over the top in some performance tests. AMD is a hardware company first and the software is there to sell the hardware. This makes perfect sense.
0 Likes
Quoting: CimerydQuoteThat's pretty good of AMD to offer it up like that, but to what end? What would be the point of having an OpenGL API that very closely mimics the Mantle API? You would then have two competing API's that are near the same.My first thought is that AMD are fighting a losing battle on both the CPU and GPU front, being quite clearly second best at both. "Helping" both OpenGL and DirectX by giving them code and direction tailored to AMD hardware will bridge that gap, and might even push them over the top in some performance tests. AMD is a hardware company first and the software is there to sell the hardware. This makes perfect sense.
I'm sure you guys will jump on me for this, but they appear to be the underpowered but of good morales on the Linux side and while it's hard to deny that the CPUs are not competitive enough, the GPUs on Windows are still competitive.
However it's these little things that make me stick with AMD. I would take an open standards advocator over more power any day.
0 Likes
QuoteMy first thought is that AMD are fighting a losing battle on both the CPU and GPU front, being quite clearly second best at both.
I would agree with the GPU side of things (especially on Linux), but with lower-end CPUs AMD clearly trumps Intel.
If you look at the FX-6300 and the i3-4130 (both priced the same), the AMD has a passmark of 6359 while the Intel has just 4816. The FX-8350 has 9024 and the i5-4440 has 6500, despite being newer and more expensive and has roughly the same score as the FX-6300 despite costing nearly twice as much. Really, for gaming AMD is the better choice, and getting the more expensive i5 and i7 CPUs (where Intel performs far better) is complete overkill unless you do heavy video editing, etc.
Even if you get one of the best Intel cards, gaming performance is almost identical to those two AMD cards. The myth that you need an i5 or i7 for gaming is simply untrue, that £70 FX-6300 will give you nearly identical gaming to a £230 i7:
https://www.youtube.com/watch?v=66BiQsOM9_M
Hopefully that doesn't come across as fanboy-ism, I have no loyalty to one or the other, just like to spend money wisely.
0 Likes
Quoting: Segata SanshiroQuoteMy first thought is that AMD are fighting a losing battle on both the CPU and GPU front, being quite clearly second best at both.I would agree with the GPU side of things (especially on Linux), but with lower-end CPUs AMD clearly trumps Intel.
I don't think its fair to say that. The Linux drivers are still pretty close on Windows for performance but what people ALWAYS fail to realise is that there are many other factors to choosing a GPU other than performance per watt or per currency.
- If a new Xorg came out tomorrow, radeon would be supporting it from day one.
- If xfce-wayland came out tomorrow, radeon would run out of the box on it.
- If nvidia came out tomorrow and stopped supporting, even if it is theoretical, there would be no obligation by them to release code or do anything.
TL;DR Preformance!=Best
And yes that was fanboyism, not for AMD but for Linux and FOSS. I couldn't care less about catalyst either, but I damn well want mesa to take over.
0 Likes
Well if they can pick what's good from Mantle, that's the great thing about FOSS. =)
Also, if in an engine a dev can just click/check "Build to Linux via Mantle" and "Build to Linux via OpenGL" then win win win =)
Also, if in an engine a dev can just click/check "Build to Linux via Mantle" and "Build to Linux via OpenGL" then win win win =)
0 Likes
Quoting: Half-ShotI don't think its fair to say that. The Linux drivers are still pretty close on Windows for performance but what people ALWAYS fail to realise is that there are many other factors to choosing a GPU other than performance per watt or per currency.
- If a new Xorg came out tomorrow, radeon would be supporting it from day one.
- If xfce-wayland came out tomorrow, radeon would run out of the box on it.
- If nvidia came out tomorrow and stopped supporting, even if it is theoretical, there would be no obligation by them to release code or do anything.
nouveau is improving in huge leaps and all amd points you mentioned are valid whole mesa stack, not just amd.
nvidia dropping binary support+nouveau getting better would just put them at almost same stage as amd blob wise. amd is just getting worse and worse in that department. i really hope this is sign of joining oss+blob initiative and that would be the reason why they are dropping the ball on linux blob right now. oss driver from amd on the other hand it's like they put it in the next gear or 2 development wise. so, i kinda hope that is it
Quoting: Half-ShotTL;DR Preformance!=Best
And yes that was fanboyism, not for AMD but for Linux and FOSS. I couldn't care less about catalyst either, but I damn well want mesa to take over.
that i couldn't agree more with. i'd gladly pay for rank or two higher card and get less performance, but open driver
0 Likes
See more from me