AMD GPU14 Tech Day Event - Sept 25'th

[Alexko]

I think people have to start thinking in terms of power consumption first. Because 20% of a CPU core may not be huge (though other algorithms may require more resources) but it's a significant amount of power, which probably dwarfs what TrueAudio requires.

And power saved is power that can be spent on extra performance (via Turbo).

 

[gamervivek]

It instantly reminded me of this interview from early 2011:

amd playing dr. jekyll and mr. hyde

Of course, programming direct-to-metal isn't for every games developer. We mentioned the possibility to Introversion's lead designer and developer Chris Delay, for example, who simply said: 'I don't want anything to do with that, but presumably it depends on what you're developing. If you're making Crysis 3 or something like that, then it may be exactly what you want.'

Indeed, Crytek's R&D technical director, Michael Glueck, said 'yes, that would appeal to us,' when he heard about Huddy's claims. However, he also seemed surprised, pointing out that 'AMD is the biggest force driving the OpenCL standard; it even released drivers to run OpenCL on SSE2 and that's kind of a way to remove low-level hardware access for simple compute tasks.'

http://www.bit-tech.net/hardware/graphics/2011/03/16/farewell-to-directx/3

 

[Davros]

How can we compare when you provide a benchmark at a resolution the ps3 is incapable of ?
plus the situation is not the same today a lot of work has been done to make dx11 leaner

edit : ninja edited by about 5 people

 

[Andrew Lauritzen]

Mantle is not for every developer. It is a low-level graphics API designed to drive the GPU in the most efficient manner. This level of access requires a bit more development effort than existing APIs to reap the rewards it provides.

Can you say whether it's going to be a public API/spec or whether it'll only be under NDA for specific folks?

More technical details about Mantle will be revealed in due course. Johan is a keynote speaker at the AMD Developer Summit (11-14 November, San Jose).

Great, yeah I look forward to hearing the details!

Thanks for the clarification Nick.

 

[CarstenS]

I am sure yesterday some people in Sony/MS camp upped their eyebrow when listened to the AMD conference. Even the trueaudio presentation could have raised suspicions.

I agree, also with the TrueAudio part.

I don´t agree. Compare a ps3 game with a PC version of the same game with a 7600GT and after that with a more powerful 7800GTX. For example: Oblivion that came out at the PS3 launch windows,

http://images.anandtech.com/reviews/video/Oblivion/oblivion-highend.png

What is then the point of a low level api and the complaints of developers about drawcalls, poor use of gpu resources and the likes?. Why Repi is so happy developing Mantle and why the R9-290X Mantle version will kill a Titan DirectX version in Battlefield 4?.

Well, apart from Oblivion, the other three descriptors in that benchmark are not met with the PS3: High Quality, 1280x1024 and HDR. Especially the latter let Oblivion performance tank on G70-class hardware.

 

[Eolirin]

I'm not sure why anyone thinks this is bad for MS or Sony. This won't have any impact on them at all. This is entirely about getting a competitive advantage out of software because there's no way for AMD to get a big enough advantage out of hardware to change their marketshare situation. And that's smart, but let's not act as if this changes any of the market dynamics that exist for consoles or OSes.

For MS, Windows PCs are dominant in the market this would apply to, and not because MS has a lock on gaming related APIs. And no one that really wants a console gaming experience is going to jump ship to PC gaming, which will still have all of it's existing issues despite this. PCs have been stronger than consoles pretty consistently and it hasn't changed the dynamics of that market. This is just more performance.

If SteamOS is successful it'll be successful regardless of what AMD is doing; having access to the same functionality as Windows isn't going to be a competitive advantage for Valve.

 

[lanek]

Well I think Anand got this wrong (with all due respect), I would think that the XB1 api is based on whatever AMD was working on, not the other way around.
I agree with Carmack MSFT or Sony might not be happy with to see some competition taking shape already when they have yet to release their new systems supposed to last for who knows how long.

Isnt Carmack was praise for low level API during all this years ? I think remember this was his main battle ... what the problem ? its not Nvidia or ?

I really well imagine Nvidia presenting Mantle, and seeing Carmack coming in the scene explaining how great it is.

 

[Dave Baumann]

http://www.tweaktown.com/news/33154...-three-4k-monitors-via-the-r9-290x/index.html

3x 4K Eyefinity and Crossfire.

 

[repi]

I am skeptical about the performance gains to be achieved from Mantle. True that the current DX driver stack is deep and adds an overhead but you would only see large gains when it is actually a bottleneck. DX11 already has deferred contexts which allows multiple CPU cores to feed the GPU using display lists.

In our experience, DX11 deferred contexts give no performance gains at all, the opposite actually, so have completely given up on it which is sad but true.
See my slide #34 here:
http://www.slideshare.net/DICEStudio/directx-11-rendering-in-battlefield-3

 

[oscarbg]

Two questions:
For me most interesting question not really clarified is Mantle will come to MacOS, really? that
would put MacOS from now currently OGL 3.2 support (DX10 more or less) (right 4.1 (partial DX11
features) is around the corner) to state of the art.. Dave Baumann can AMD is interested in true
cross-platform support for Mantle i.e. Win,Lin+MAC..

Also other question is about StreamOS:
Do you expect solid StreamOS Intel GPU support (for Intel only StreamBoxes (i.e. without AMD or NV GPUs)) to ignite interest in Intel to have OGL Intel Win binary drivers ported to Linux now Win OGL drivers offer GL 4.2+key parts of GL 4.3 like compute shaders (i.e. full DX11 parity).. in contrast current Intel Linux efforts have Mesa at GL 3.3 this year perhaps and not full 4.0 even in years to come..

 

[eastmen]

I'd expect Mantle to help PC pull away from (or keep up with, depending on your hardware) consoles faster. Of course, the consoles still have some CPU/GPU integration advantages, so there still be some room for optimisation not immediately available in the PC space yet.

the R9 280xt or whatever the top line chip is already far and away more powerful than next gen systems and will most likely beat them out to customers. Next year we will see the first 20nm gpus and it will explode the gap.

We should finally get a new bulldozer over the next year and we will get broadwel . So I think by the end of 2014 any advantage a console might have will be gone.

 

[liolio]

Isnt Carmack was praise for low level API during all this years ? I think remember this was his main battle ... what the problem ? its not Nvidia or ?

I really well imagine Nvidia presenting Mantle, and seeing Carmack coming in the scene explaining how great it is.

I don't get your post, I just pointed out that he said that MSFT and Sony may not be that happy about the hardware vendors and content creators showing some willingness to break away from the constrained (when not closed) environments MSFT and SOny want them to be locked in.

How that anything to do with Nvidia? Or some good relation JC would hae with Nvidia?

 

[Raqia]

OpenGL is fully out of the domain of Windows and wholly in the hands of the IHV's, Mantle will be no different in this respect. Memory re-write was already implemented many releases ago.

Thanks for the response! Is this API going to be open? It could be pointless since it must be closely tied to AMD hardware and synergistic w/ AMD tools written for consoles so w/o a tight mapping of commands and functions to hardware level instructions, using mantle could be moot. Looking forward to it though.

 

[Kaotik]

Thanks for the response! Is this API going to be open? It could be pointless since it must be closely tied to AMD hardware and synergistic w/ AMD tools written for consoles so w/o a tight mapping of commands and functions to hardware level instructions, using mantle could be moot. Looking forward to it though.

From TechSpot (and no doubt gazillion other sources)

We've been told at the GPU14 Tech Day event that the Mantle API is open, so theoretically Nvidia could purpose the technology in their GPUs.

 

[Andrew Lauritzen]

I don't believe it, or rather I'm sure there was a miscommunication/misinterpretation involved. Even if AMD intends to publicly disclose the API/specification, they have made it very clear that it is designed to target GCN... not older AMD architectures, definitely not other hardware vendor's architectures. Besides, if they were really intending to make it portable to other hardware there would be no reason to not propose it to Microsoft/Khronos and try to do it through changes to DX/GL first. To my knowledge, they haven't done anything of the sort.

Feel free to correct me if I'm wrong Dave, but I doubt it.

 

[Dave Baumann]

To your knowledge.

We are taking the first step with Mantle, albeit one that already has direct target implementations with developers. HSA started life as something very specific to AMD's Fusion direction and now has fairly broad industry support.

 

[Andrew Lauritzen]

I'm not saying this isn't something that couldn't develop into broader industry support; I'm responding specifically to the claims that someone will be able to take Mantle as it is today (or rather, in December) and support it on non-GCN hardware for the purposes of running BF4's new path or similar. That I seriously doubt.

In terms of affecting future APIs, absolutely these things are always evolving and I hope that's a direction AMD intends to pursue. But the resulting APIs would still be evolutions of DirectX/GL, even if a clean break is needed. In my opinion that's the ideal direction to go here unless it's proven than *any* abstraction/portability makes too large a performance sacrifice. That has yet to be established though.

I guess a lot of this is going to remain fairly unclear until November when more details are given.

 

[Alexko]

To your knowledge.

We are taking the first step with Mantle, albeit one that already has direct target implementations with developers. HSA started life as something very specific to AMD's Fusion direction and now has fairly broad industry support.

Thanks for the info.

I really hope Mantle gets the same kind of support. If NVIDIA adopted it, it would be fantastic news for the PC gaming world.

 

[Megadrive1988]

PS3 had a different architecture all together with a very powerful CPU, it also had an almost top of the line GPU (at that time). this is not the case anymore .

No not really.

PS3 came out in late 2006. The RSX GPU was a cut down version of the NV47 / G70. As mentioned it had half the ROPs and memory bus width of the high-end PC versions.

On top of that, Nvidia had already launched the G80 for PCs so PS3 most certainly did not have an almost top of the line GPU at the time it launched.

 

[Arwin]

I've always wondered if it is really half, or if the other half was just hooked up to the CPU memory pool.