Will AMD's Mantle revolutionise PC gaming? *spawn

[Scott_Arm]

Apparently a single 290X was driving the 5760x1080 live play of BF4 Angry Sea and no one said it was running in Mantle mode.

So, erm, it seems PC gamers don't need Mantle

I want to see how much it helps the low and mid range GCN cards. That's the real win. If a lot more people can suddenly run Battlefield 4 at respectable framerates and graphical settings, it'll be a big selling point.

 

[Silent_Buddha]

Even more exciting is to think what Mantle could conceivably meant for low power Windows laptop/tablet gaming. Perhaps you won't have to have a desktop replacement laptop that gets at best 1-2 hours on battery life for good gaming performance.

In other words, I wonder what this could do for a low power Windows laptop/tablet that gets 8-10 hours of battery life. Obviously lower when gaming, but perhaps gaming will actually become pleasant on such a low power device with something like Mantle.

Regards,
SB

 

[Babel-17]

I'm not technically minded but I'm wondering if Microsoft is also learning from the combined effort behind the Xbox One.

Maybe they are looking to implement some of that in future operating systems.

This is exciting stuff. More free cpu cycles available to developers could really be an incentive to make games with beefy AI.

 

[eastmen]

Even more exciting is to think what Mantle could conceivably meant for low power Windows laptop/tablet gaming. Perhaps you won't have to have a desktop replacement laptop that gets at best 1-2 hours on battery life for good gaming performance.

In other words, I wonder what this could do for a low power Windows laptop/tablet that gets 8-10 hours of battery life. Obviously lower when gaming, but perhaps gaming will actually become pleasant on such a low power device with something like Mantle.

Regards,
SB

wouldn't amd actually need to get window's tablets out that have gcn hardware ?

 

[Jawed]

I'm not technically minded but I'm wondering if Microsoft is also learning from the combined effort behind the Xbox One.

Maybe they are looking to implement some of that in future operating systems.

They already had that opportunity with XB360.

It would have influenced features in D3D10/11, but it's had pretty much zero effect in terms of improving the API by reducing its overheads.

The last hope would appear to be a true break API, if D3D12 threw out everything vaguely fatty and went Mantle style. D3D11 and OpenGL can remain for the devs who are too lazy (prolly only OpenGL, truthfully).

Microsoft can tell NVidia: do this or fuck off. Which IHV has been slowing down D3D since DX8 the most: AMD (ATI) or NVidia. We can have a poll: but, I don't think there's much danger of NVidia losing that.

NVidia has its focus on mobile, making developers in that space happy with proprietary extensions to mobile graphics and writing code for them. Its doing everything it can to own mobile gaming, a second shot at the strategy that failed in PC's gaming space.

If NVidia gives a damn about PC gaming it'll do its own Mantle in a bid to make Microsoft irrelevant in PC gaming.

But I can't help thinking it's distracted by Shield and CUDA. NVidia would prolly rather we had Shield emulators on PC that locked out Microsoft and AMD. Wait, did I say that? Shit I don't want to give them that idea.

This is exciting stuff. More free cpu cycles available to developers could really be an incentive to make games with beefy AI.

There's already a vast number of free CPU cycles

 

[Davros]

One problem amd may have is nvidia come out with its own api, do it better, spend a hell of a lot more on devrel maybe even paying devs(cash or work they offered to add physx support to titles for free) also I could imagine nvidia making devs sign exclusive agreements.

 

[BRiT]

One problem amd may have is nvidia come out with its own api,

Hasn't Nvidia already released their Bindless OpenGL extensions back in 2010 which seems to do much of what Mantle is claiming to do (remove the overhead of draw calls)?

 

[liquidboy]

VirtualGPU

Lets say hypothetically that the AnanandTech article on "Understanding AMD’s Mantle: A Low-Level Graphics API For GCN" has a correct interpretation of Mantle, seeing as @AMDRadeon twitter account itself recommends it as a read. Maybe not 100% correct, as we'll never get that confirmation.

Now in that article there are 2 interesting comments that I've come across from someone that sounds like they know what they're talking about with regards to "Mantle" . And it meshes with what I've been thinking about with regards to Xbox One..


[note: you may want to go to the comment(s) in the article themselves and read the preceding parent question that resulted in the comments. This may give you more sense of what Jaybus is trying to say]

comment 1:

Huh? OpenGL is a very high level API. In fact, it is a general purpose graphics API that is completely abstracted from the hardware. Yes, Mantle is an API, but its functions have a 1:1 correspondence with a hardware function in the same manner that the _InterlockedExchange() intrinsic function maps to the xchg4 opcode on x86 hardware. It is mostly like intrinsics, although for GCN hardware instead of x86. Register assignment, etc. is handled by the compiler (or the API function itself is coded in assembler).

comment 2:

Wait. I think the question is could DX and OpenGL drivers be written to use Mantle rather than direct hardware access. The answer is yes. Mantle is providing a virtual GPU that is a direct 1:1 mapping to hardware functions. It might not be quite as fast, but it should be pretty close because it wouldn't need make any more mode switches than it would for direct hardware access. The benefit would be in driver development. Once a particular OpenGL or D3D version was implemented using Mantle, it would work for any AMD GCN hardware well into the future. It would give AMD a set of stable, optimized D3D and OpenGL drivers that would work with any future GCN devices for quite some time and require far less code maintenance.

The whole idea of removing the hardware method calls from the DirectX drivers, and whatever other drivers in user/kernel mode, and leveraging MANTLE api calls makes sense to me.

And if it can be done in a way that's near free of perf (I know theres a cost involved BUT doesn't sound like its much) .. And you get this "compatibility" story and "cross platform pc" play well into the future ...

Sounds like a great thing to have!

 

[Brad Grenz]

I'm not sure I understand what your point is? He's saying that you could build DX or OGL drivers on top of Mantle rather than a GPU's native instructions. Obviously that's possible, just the way those miniGL drivers for 3Dfx cards were written in Glide, or a Glide wrapper used today is written in DX11 or OGL. If you want to badly enough you can always insert a translation layer, I'm just not seeing a major benefit. The quote from the commenter above mentions the only one I can think of, smoothing over driver development for AMD's future products, but it doesn't make DX or OGL applications faster, nor do I see any relationship to Xbox One.

 

[Davros]

Hasn't Nvidia already released their Bindless OpenGL extensions back in 2010 which seems to do much of what Mantle is claiming to do (remove the overhead of draw calls)?

Yes, but I can only think of 2 games that use it divinity Dragon Commander and possibly rage (was rage too early)
oh maybe brink

 

[liquidboy]

I'm not sure I understand what your point is? .... nor do I see any relationship to Xbox One.

I have a picture in my head of how the Xbox One Software is setup at the OS level (HyperV + GameOS + AppOS) and how its meant to access HW (SoC+other custom silicon) ...

I have an idea of how it all hangs together, it involves a Mantle like vGPU layer plus the Mono(d3d) + other drivers work ...

Regardless I found those comments interesting, others may not!

 

[liquidboy]

but it doesn't make DX or OGL applications faster, ...

No one ever assumed this ...

 

[Dominik D]

The whole idea of removing the hardware method calls from the DirectX drivers, and whatever other drivers in user/kernel mode, and leveraging MANTLE api calls makes sense to me.

That's because you've never written a GPU driver. "Virtual GPUs" is what effectively abstractions on the WDDM level give you. But that comes at a cost which no API is going to cut (because "virtual GPUs" are about sharing resources between many consumers, and this comes at the cost of context store a.k.a. pre-emption, which is used by WDDM drivers today as efficiently as possible).

Also I think people are reading WAAAAY too much into the fact that Radeon retweeted Anand's article. If I were responsible for PR, I'd retweet every piece of positive info about my company, regardless of whether it was entirely correct, or not.

 

[Rodéric]

Didn't check, was Mantle announced for Linux ?
With SteamOS coming I'm wondering, a cross OS, low level, high performance, developer empowering API would be very welcome...

 

[Dominik D]

It wasn't.

 

[willardjuice]

It wasn't.

Well they said multi-platform no (launching on Windows first)? I took that to mean Windows/Linux/OSX eventually.

 

[Dominik D]

Sure, but it wasn't announced for Linux, was it?

 

[Brad Grenz]

Mantle's second supported platform will be BeOS. Then WebOS. Then OS/2. Then FreeBSD. Then Linux.

 

[Dominik D]

Right, it's not like Apple could switch to AMD. Also typically when things like these are announced we get the "this is bound to a single platform!" memo as a sign of good PR. Plus: last time I checked announced meant announced, not hinted or suggested.

 

[grndzro]

One problem amd may have is nvidia come out with its own api, do it better, spend a hell of a lot more on devrel maybe even paying devs(cash or work they offered to add physx support to titles for free) also I could imagine nvidia making devs sign exclusive agreements.

The difference is AMD's Mantle has more developer support than Nvidia's after 1 month vs several years. NV will eventually support Mantle IMO. They would be incredibly stupid not to take advantage of AMD's inroads.