Erm...why? How does that make it different from, say, CUDA?
The whole benefit of Mantle dies if it is to be cross ISA compatible. Then we can just as well stay with OpenGL and DirectX.
Mantle is a direct "to this specific metal" patch to games. Think 3DNow! patches all over again, allbeit the next CPU may not support 3DNow! but 3DNow! II .. So you need a new patch. There is good and there is bad. You'll need the patch! Without it you're screwed on an APU. Get the patch? - You gain awsome performance - Rock'n Roll dancing with the stars.
But isn't that like everything else? How is that any different than needing OGL 4 when you have OGL 3? Or DX 11.1 when you have 11? Or when you are running Flash 10 and you need 11.8? Or Java 7? Or Android 4? Or iOS 7?
What is this "patch", why is it different? How is it different than what we already have?
And if you need new hardware?(Like everything else...) What's so bad about it?
Aren't we used to this?
And what would happen, if it doesn't have to?The whole benefit of Mantle dies if it is to be cross ISA compatible.
And what would happen, if it doesn't have to?
GCN seems to be arch independent to some extent, since it's a scalar ISA part (ctrl flow, scalar ops) + a vector ISA part (512b SIMD).
Improvements to the overall arch could happen w/o changing existing ISA definitions, but just extending them, which would mean backwards compatibility to existing code.
The same way it worked for x86.
It is different cause you have one API taking care of everything, all the legacy. You can move it around but the ah heck is still gonna be in there somewhere.
Compare it to x86 and all the talk about how much better we would be with a more modern ISA. They are right. The story is that a couple of IBM'ers hacked up x86 in two weeks and that is still what we have today .. due to legacy. If we were able to shed that legacy easily we would all be running some iteration of itanium by now (wish we were). Point being, being dead hard on backwards compatability and on multiple platforms will eventually crawl you down to a slow. Back when DX was thought up the idea of an APU didnt exist so how could they anticipate for it?
There is good and there is bad. Yes we should consolidate api's for all the right reasons and we should also abandon the same apis for another set of good reasons. While Mantle isnt event in revision 1.0 ... i really dont understand all the judgement. It could be great.
I am quite aware of your logic. So the day GCN goes out, Mantle goes out.
You realize that the R600 architecture which launched in Xbox 360 and HD 2900 XT (May 2007) powered all GPUs till HD 6970 (Dec 2011). remember VLIW4 was just a tweak to the same VLIW5 architecture.
But isn't that like everything else? How is that any different than needing OGL 4 when you have OGL 3? Or DX 11.1 when you have 11? Or when you are running Flash 10 and you need 11.8? Or Java 7? Or Android 4? Or iOS 7?
What is this "patch", why is it different? How is it different than what we already have?
And if you need new hardware?(Like everything else...) What's so bad about it?
Aren't we used to this?[/QUOTE
Well, the negative I see is that it is hardware specific. It not like everyone can benefit from it. That is just unfortunate if you have nVidia, but I fear even worse is that it could slow improvements in direct x and lead developers to make poorly optimized Dx code paths. After all, there is a limited amount of resources a developer can devote to a PC port, and I am sure amd will be "encouraging" developers in GE games to devote as many resources to mantle as possible. And if nVidia cones out with their own response, that is even more to code for.
And I an not as cavilair as you seen to be about getting new hardware.
No, we can't agree on this.This might be possible, but that means a lot of development and validation effort from AMD, and I think we can agree that this is far more effort than the company has been giving on its entire life. AMD usually goes to a "we build hardware and software will catch up later".
Is that comparable?So can we expect AMD to give the same level of support and thought that Intel gives to x86? IMO this is a risky proposition.
Well, the negative I see is that it is hardware specific. It not like everyone can benefit from it. That is just unfortunate if you have nVidia, but I fear even worse is that it could slow improvements in direct x and lead developers to make poorly optimized Dx code paths. After all, there is a limited amount of resources a developer can devote to a PC port, and I am sure amd will be "encouraging" developers in GE games to devote as many resources to mantle as possible. And if nVidia cones out with their own response, that is even more to code for.
If you're discussing mantle, first, with regards to DirectX. It took MS nearly 10 years to evolve from DX9 to DX11. The technology for tessellation was around in 2006 but it took them forever to catch up - so the suggestion that DX will slow? Sorry, the truth is that MS are the ones that are slow, period.
Secondly, I don't really feel Mantle will require more developer resources - what you're unaware of is that AMD is targetting Mantle at games engines, not specific game developers. The fact of the matter is, there are 2-3 game engines that power 85% of games across all platforms, and Unreal Engine is the biggest; AMD is trying to get Mantle functionality into these engines - If the engine supports Mantle, any developer using that engine will have Mantle access by simply licensing it. This does not require more developer resources. Frostbite 3 is the first, Crytek has also expressed optimism regarding Mantle, and UE3/4 will be the next choice. AMD is not going to have a situation where they talk to 3000 developers and waste time doing this. They do not have to. All they have to do is get engine functionality - with Frostbite 3 powering nearly every EA game for the next 1-2 years, every developer using Frostbite 3 will have Mantle access. This includes the upcoming NFS and Dragon Age: Inquisition games.
MS will not provide low level access
I have no opinion on the ARM based SOCs that AMD is creating, though, I don't see anything there that could differentiate AMD from the many other ARM SOC vendors. If you're discussing something other than Mantle, my mistake.
You realize that the R600 architecture which launched in Xbox 360 and HD 2900 XT (May 2007) powered all GPUs till HD 6970 (Dec 2011). remember VLIW4 was just a tweak to the same VLIW5 architecture.
similarly GCN has a long life ahead and as the architecture which powers the next gen consoles PS4 / XB1 it should have no problem. GCN 2.0 will drive the upcoming Hawaii chip and there will be future iterations which maintain the core architectural compatibility with the original GCN 1.0. Obviously the future products will be super sets and support higher efficiency, more advanced functionality and bring higher performance. The GCN architecture and its enhancements should be around till 2016 or even further.
here is the same question you raise asked by hardware.fr to Raja koduri
http://www.hardware.fr/focus/89/amd-mantle-interview-raja-koduri.html
"Could we imagine that in a couple of years, a future architecture won't be able to run the first Mantle games ? Or did you design Mantle to be forward compatible ? There's always this kind of tradeoff with a lower level access
RK - Those are all great questions but Frankly we'll see how it goes. At the end of the day forward compatibility and backward compatibility are important aspects but if they're getting in the way of solving a problem at a given point of time, if they're getting in the way of exposing something that the new hardware is capable of that makes the game be hundred times more realistic, we have to be practical about it and thats how we move things forward. We move technology forward and at some point of time we have to say "out with the old compatibility", and move forward. If not you get stuck. "
Sandy, ivy and haswell are all tweaks on the foundation laid down by Nehalem. higher IPC, newer instruction sets like AVX / AVX 2. But they are not radical departures from the Nehalem core. all code which was written for nehalem works on haswell . but the vice versa need not be true. maintaining such a compatibility for 5 - 6 years on the future GCN GPUs is possible.
Is that a FUD attempt?Yea, and if getting rid of compatibility with legacy code just happens to screw our competitor, that is a nice side benefit.
Erm, i'm confused, but I simply hopped in to the last page of this thread.
If you're discussing mantle, first, with regards to DirectX. It took MS nearly 10 years to evolve from DX9 to DX11. The technology for tessellation was around in 2006 but it took them forever to catch up - so the suggestion that DX will slow? Sorry, the truth is that MS are the ones that are slow, period.
Secondly, I don't really feel Mantle will require more developer resources - what you're unaware of is that AMD is targetting Mantle at games engines, not specific game developers. The fact of the matter is, there are 2-3 game engines that power 85% of games across all platforms, and Unreal Engine is the biggest; AMD is trying to get Mantle functionality into these engines - If the engine supports Mantle, any developer using that engine will have Mantle access by simply licensing it. This does not require more developer resources. Frostbite 3 is the first, Crytek has also expressed optimism regarding Mantle, and UE3/4 will be the next choice. AMD is not going to have a situation where they talk to 3000 developers and waste time doing this. They do not have to. All they have to do is get engine functionality - with Frostbite 3 powering nearly every EA game for the next 1-2 years, every developer using Frostbite 3 will have Mantle access. This includes the upcoming NFS and Dragon Age: Inquisition games.
While the performance gains associated with Mantle remain to be seen, I think it's a great idea. Fact of the matter is, highly respected developers have lamented the API situation with DirectX because MS is lazy, period. MS will not provide low level access - consider the fact that PC dGPUs are 25 times more powerful than current consoles. Yet we can only harness a TINY fraction of that. If a new situation comes along where that can be partially rectified, i'm all for it. DirectX made sense in 1998 because there were 10 proprietary 3d chips with no standards - such as the rendition verite, s3 virge, 3dfx, among many others. Now we still have directX for catch all functionality, while DirectX is lagging in development and does not let us harness the potential of our hardware.
I don't know how this ties in to the topic, but this is my view of Mantle. It's a good thing. I have no opinion on the ARM based SOCs that AMD is creating, though, I don't see anything there that could differentiate AMD from the many other ARM SOC vendors. If you're discussing something other than Mantle, my mistake.
Erm, i'm confused, but I simply hopped in to the last page of this thread.
If you're discussing mantle, first, with regards to DirectX. It took MS nearly 10 years to evolve from DX9 to DX11. The technology for tessellation was around in 2006 but it took them forever to catch up - so the suggestion that DX will slow? Sorry, the truth is that MS are the ones that are slow, period.
Secondly, I don't really feel Mantle will require more developer resources - what you're unaware of is that AMD is targetting Mantle at games engines, not specific game developers. The fact of the matter is, there are 2-3 game engines that power 85% of games across all platforms, and Unreal Engine is the biggest; AMD is trying to get Mantle functionality into these engines - If the engine supports Mantle, any developer using that engine will have Mantle access by simply licensing it. This does not require more developer resources. Frostbite 3 is the first, Crytek has also expressed optimism regarding Mantle, and UE3/4 will be the next choice. AMD is not going to have a situation where they talk to 3000 developers and waste time doing this. They do not have to. All they have to do is get engine functionality - with Frostbite 3 powering nearly every EA game for the next 1-2 years, every developer using Frostbite 3 will have Mantle access. This includes the upcoming NFS and Dragon Age: Inquisition games.
While the performance gains associated with Mantle remain to be seen, I think it's a great idea. Fact of the matter is, highly respected developers have lamented the API situation with DirectX because MS is lazy, period. MS will not provide low level access - consider the fact that PC dGPUs are 25 times more powerful than current consoles. Yet we can only harness a TINY fraction of that. If a new situation comes along where that can be partially rectified, i'm all for it. DirectX made sense in 1998 because there were 10 proprietary 3d chips with no standards - such as the rendition verite, s3 virge, 3dfx, among many others. Now we still have directX for catch all functionality, while DirectX is lagging in development and does not let us harness the potential of our hardware.
I don't know how this ties in to the topic, but this is my view of Mantle. It's a good thing. I have no opinion on the ARM based SOCs that AMD is creating, though, I don't see anything there that could differentiate AMD from the many other ARM SOC vendors. If you're discussing something other than Mantle, my mistake.
/me wonders what this has to do with the original topic?
Well, I'm gonna guess here. The benefit of GCN in an ARM based SoC?
Without Mantle, I guess it would be just like a Tegra, nothing.
None of the engines Mantle is targeting are removing DX11. That is, put simply, an incorrect assessment of the situation. AMD is not targeting single developers. They are targeting engines, and none of these engines will lose DX11. Nvidia will still thrive because they make awesome products with great software and great DX11 performance.
This is simply AMD's value add, and we don't know how well Mantle even performs yet. It could be very worthwhile, or not, but I applaud AMD's effort to match nvidia's efforts to create value added features. Nvidia has physx, TXAA, and all that stuff and it works only on NV hardware. Mantle is AMD's value add. Period. Stating that DX11 will go away because of Mantle is just not correct - all of the engines that Mantle is directed to will still have excellent DX11 implementations. And nvidia will still be nvidia - they will release awesome products with great software just as they always have.
I dont see it as the same as physX really. PhysX is a "value added" function because the developer has to optimize to the same extent for DX whether they use physX or not.