The AMD Mantle Thread

[Lonbjerg]

I suspect that could be a reason why AMD decided to come up with Mantle in the first place, as it removes the need to create multithreaded drivers..

As game engines become more and more threaded, NVidia will gain a strong advantage as their drivers are not only highly threaded, but they support DX11's driver multithreading; which will surely continue to evolve.

The PClab.pl review inadvertently showed that AMD has a serious weakness when it comes to multicore support within their drivers.. And since they haven't bothered to implement DX11 multithreading, it's probably safe to assume they either don't have the resources, or the level of competence to do so..

It makes it even more fun...when people tlak about AMD being "fully" DX11.2 compliant...when their driver isn't multithreaded.

And I suspect you are correct...AMD has been real long in the tooth making a DX11 multithreaded driver.
One can wonder why AMD's tries Mantle..and not just fix the driver first.

 

[Gloomy]

It makes it even more fun...when people tlak about AMD being "fully" DX11.2 compliant...when their driver isn't multithreaded.

And I suspect you are correct...AMD has been real long in the tooth making a DX11 multithreaded driver.
One can wonder why AMD's tries Mantle..and not just fix the driver first.

Is this a joke post? You are joking, right? :|

 

[Lonbjerg]

Is this a joke post? You are joking, right? :|

You don know either that AMD's DX11 dosn't support multithreading and never has?
Is this really forgotten?

http://forums.anandtech.com/showpost.php?p=31520674&postcount=28

I think you should read a bit up (using that link) on what multithreading in the DX11 drivers does....Try finding the phrase: "Traditionally, rendering is a very serial process..."

Mantle and DX11 multithreading sounds to offer some similar "benefits"...one can wonder why AMD dosn't support the function...for EVEN mor added performance?

Does it look like I am kidding now?
Can you find more vaild arguments?

 

[zlatan]

It is important to understand that not AMD want to create a low-level access. Developers have been asking for a better API for years.

I always prefer standards, but when I have to tell the artists to optimize batch count ... well they don't like me much, and I can understand that.

Also there are many people asking these questions: Why modern games lack of innovation? Is it because the designers don't have new ideas? Well no. They have plenty of awesome ideas. But when they share these to the programers, than ... sure we will say that's a cool idea, but it isn't posibble on the actual APIs.

Even if Mantle will not succeed, it will prove that something is seriously wrong with the standard programing interfaces.

 

[BallaTheFeared]

It is important to understand that not AMD want to create a low-level access. Developers have been asking for a better API for years.

I'm sure when they were asking for it they were hoping it would work on more than the 5% of the market GCN represents.

 

[blastingcap]

I'm sure when they were asking for it they were hoping it would work on more than the 5% of the market GCN represents.

Nice try. GCN is in APUs and the latest AMD GPUs, which total more than 5% of the market.

 

[DarkKnightDude]

Consoles also count as GCN tech don't they?

That Oxide demo looked impressive and they said that's with zero optimization .

 

[psoomah]

You don know either that AMD's DX11 dosn't support multithreading and never has?
Is this really forgotten?

http://forums.anandtech.com/showpost.php?p=31520674&postcount=28

I think you should read a bit up (using that link) on what multithreading in the DX11 drivers does....Try finding the phrase: "Traditionally, rendering is a very serial process..."

Mantle and DX11 multithreading sounds to offer some similar "benefits"...one can wonder why AMD dosn't support the function...for EVEN mor added performance?

Does it look like I am kidding now?
Can you find more vaild arguments?

While I would point out your referenced link is 2 years 8 months old, and therefore way out of date, I also noticed this fascinating little gem tucked away near the end of Ryan Smith's article:

"Coincidentally, last month's interview with AMD's Richard Huddy at Bit-Tech also has a lot in common with this. AMD says DX11 multi-threaded rendering can double object/draw-call throughput, and they want to go well beyond that by bypassing the DX11 API."

This from 16 March, 2011. Looks like AMD had a Mantle twinkle in it's eye from shortly after they won the PS4 and Xbox One contracts.

Johan may have technically been the first to 'formally' approach AMD with the proposal to implement a close to the metal API, but methinks he might have been nudged on the down low the time was ripe from one of the AMD collaborative team members.

 

[blastingcap]

Consoles also count as GCN tech don't they?

That Oxide demo looked impressive and they said that's with zero optimization .

I left out consoles because they have their own low-level APIs... so they aren't really in the same market space as discrete PC video cards.

And yes, I liked the demo as well despite its lack of optimization. I think Mantle is promising, and even if it never goes anywhere, at least it lights a fire under Microsoft to improve DirectX faster.

 

[Lonbjerg]

While I would point out your referenced link is 2 years 8 months old, and therefore way out of date, I also noticed this fascinating little gem tucked away near the end of Ryan Smith's article:

"Coincidentally, last month's interview with AMD's Richard Huddy at Bit-Tech also has a lot in common with this. AMD says DX11 multi-threaded rendering can double object/draw-call throughput, and they want to go well beyond that by bypassing the DX11 API."

This from 16 March, 2011. Looks like AMD had a Mantle twinkle in it's eye from shortly after they won the PS4 and Xbox One contracts.

Johan may have technically been the first to 'formally' approach AMD with the proposal to implement a close to the metal API, but methinks he might have been nudged on the down low the time was ripe from one of the AMD collaborative team members.

So AMD has a multithreaded DX11 driver now?
If not...you just posted a lot of words to say nothing.
In my book AMD is years behind delivering a DX11 multithreaded driver...if you want to talk about "outdated".

 

[psoomah]

I left out consoles because they have their own low-level APIs... so they aren't really in the same market space as discrete PC video cards.

And yes, I liked the demo as well despite its lack of optimization. I think Mantle is promising, and even if it never goes anywhere, at least it lights a fire under Microsoft to improve DirectX faster.

Microsoft is requiring the developers to use it's Xbox One custom DX API instead of direct to the metal PS4 style for a reason and I doubt it's to benefit PC gaming as a whole. Giving Microsoft's Xbox Store PC games an edge against the 'competition', every other distribution channel, seems logical.

I'm seeing Mantle being somewhat of a tidal wave Microsoft will eventually realize it cannot successfully turn back or block and Microsoft will let the developers program to the metal on the Xbox One. Why bother trying to leverage it's custom API into it's locked down PC gaming store if a substantially better PC version will still be available from other vendors and insisting on developers using the API layer further widens the hardware based performance gap the PS4 already enjoys. It's a lose lose situation.

That said it is certainly possible Microsoft, Intel and Nvidia, all of whom are disadvantaged by Mantle, might join together to propose and push an alternate solution.

 

[psoomah]

So AMD has a multithreaded DX11 driver now?
If not...you just posted a lot of words to say nothing.
In my book AMD is years behind delivering a DX11 multithreaded driver...if you want to talk about "outdated".

Ryan Smith: "Anyhow, as far as I know, AMD does not currently offer fully support for multi-threaded rendering (I don't have an AMD card plugged in right now to run the DX Caps Viewer against). I'm not sure where they are on it, though I doubt they're very far behind."

You referenced a 2 1/2 year old article to support your point, got anything current that supports your contention AMD is still behind on DX11 multi-threading?

 

[bystander36]

Ryan Smith: "Anyhow, as far as I know, AMD does not currently offer fully support for multi-threaded rendering (I don't have an AMD card plugged in right now to run the DX Caps Viewer against). I'm not sure where they are on it, though I doubt they're very far behind."

Very logical conclusion. I too doubt they were very far behind.

"In my book" means ... what exactly? Your imagination? Your donkey? You referenced a 2 1/2 year old article to support your point, got anything remotely current that supports your contention AMD is still behind on DX11 multi-threading?

Until they support DX11 multi-threading, they don't. Until there is any news that they do, or the option is available for AMD cards, it doesn't work.

 

[blastingcap]

Apparently still no DX11 MTR support if the Rage3D discussion is anything to go by: http://www.rage3d.com/board/showthread.php?t=33976408&page=7

 

[psoomah]

http://www.bit-tech.net/hardware/graphics/2011/03/16/farewell-to-directx/1

From March of 2011.

"It's funny,' says AMD's worldwide developer relations manager of its GPU division, Richard Huddy. 'We often have at least ten times as much horsepower as an Xbox 360 or a PS3 in a high-end graphics card, yet it's very clear that the games don't look ten times as good. To a significant extent, that's because, one way or another, for good reasons and bad - mostly good, DirectX is getting in the way.' Huddy says that one of the most common requests he gets from game developers is: 'Make the API go away."

"Either way, it looks as though DirectX's future as the primary gateway to PC graphics hardware is no longer 100 per cent assured, especially when it comes to cutting edge graphics."

DX was already in AMD's gun sights 2 1/2 years ago.

 

[psoomah]

Apparently still no DX11 MTR support if the Rage3D discussion is anything to go by: http://www.rage3d.com/board/showthread.php?t=33976408&page=7

Thanks. Very informative thread.

 

[3DVagabond]

I'd have to disagree on the statement that it was a "dangerous" gamble on AMD's part. I think AMD more or less saw how effective the value adds and features worked out for nvidia and wanted to emulate that somewhat, thus we have Mantle being AMD's value add. If Mantle is great, or good, or even half way decent. It will help sell their cards, period. Now we'll have debates forever about how good or bad Physx or TXAA are, but the bottom line is that there's a niche of users out there that buy nvidia cards because they enjoy the nvidia specific features. There are definitely a lot of people who buy nvidia because they say...."well I want g-sync next year. I like physx. Etc. Regardless of how someone else might NOT value it, it helps nvidia sell cards. So now AMD has that AMD specific feature or "killer app" perhaps that will help AMD drive their own sales. So they're basically doing the same thing here iMO.

We don't know how Mantle will perform yet. Like I said, it could be great or it could be just semi decent. But it will help AMD sell their GPUs regardless. I think it was the smart play on AMD's part. We don't even know how good or decent Mantle is yet and here we are talking about it 119 pages later. It created a buzz, that's for sure.

I really hope they continue down that path because I see some of the aftermarket 290X reviews and their performance is good and where it needs to be. (and without the reference issues) So they're good on that front. Software wise, they're not as good as nvidia. Yet. I would throw my personal opinion out there that nvidia is just hands down superior in terms of adding polished software features on a regular basis (for instance, TXAA, adaptive vsync, driver forced AO, shadow play, etc) So they're emulating nvidia in terms of creating value adds, hopefully they can push a little harder and get the software side up to snuff. That's the next move AMD needs to make. More software features - the last feature I can remember AMD adding was MLAA years and years ago. They need to pick up the pace on that kinda stuff. Then we all win through competition, regardless of which brand one prefers. So I guess, Mantle is the start of where AMD is headed. Hopefully they stay on the right track.

What do all of these nVidia marketing bullet points have to do with Mantle?

 

[Lonbjerg]

How can it be forgotten is a mystery too me.

Strange times when facts are forgotten...and empty PR without number as hailed.



But here we are...years later...no multithreaded DX 11 driver...but a lot of talk about "mantle"...that does the same thing...makes you wonder.

 

[3DVagabond]

AMD has no future developing CPUs. It is dead x86 wise and will be just one more in the ARM pack. The APU side of things isn't going well with Intel pushing harder every year and look at what happened with Nvidia when they tried to enter the ARM market. AMD just can't sustain the damage Nvidia is taking with Tegra and the status quo in the desktop isn't any better with the CPU division losing money every quarter.

I really don't know what kind of killer deal Apple got from AMD to include FirePros in the latest Mac Pro but AMD situation there is even worse than desktop graphics or processors with close to 0% market penetration.

Desktop graphics? 30% market share with pathetic margins and profits.

AMD in the server market? Don't make me laugh.

That's the sad state of AMD and the only thing that can save it is becoming the reference in desktop gaming since everything else failed already. That would lead to more GPU and APU sales as leaving things like this will just go worse with time.

I'm curious to see what the console sales does for the CPU division's bottom line. I'd venture a guess that all of the OoS for their high end GPU models is because they are selling tons of cards rather than because they simply don't have availability. I don't think it's all gloom and doom as you are believing.

 

[3DVagabond]

One thing that I think is really convincing in that Oxide presentation is the progression of draw calls over time in relation to GPU performance. When in 2007 we had 500Gflop/s and 3000 draw calls and today we have like 6000 GFlop/s and 5000 draw calls it really shows how much those two aspects have widened over time.

The profiles also clearly show the huge amount of CPU time that is wasted with the single threaded rendering of games via DX. Admittedly its not quite as wasted on Nvidia's drivers since it has multithreaded aspects but still the game thread is still being held up by calling into DirectX.

So for a game genre that is draw call limited (RTS and limited by units and effects for those units) it seems apparent that Mantle is going to be a huge win, that 3x performance is possible, just because the games are regularly CPU limited. Moving that bottleneck back to the GPU allows a lot more to happen in game.

A game like Battlefield 4 is probably suboptimal for Mantle actually because it isn't really draw call limited right now, but it might be interesting to see just how much more than can potentially do with it. I want gsync, I want TruAudio and I want Mantle all on 4k already. I am still bummed that I have to choose between these options. They all seem to be fundamentally great enhancements.

Yes, not good that this is the way things are turning out. Unfortunately I think it will get worse instead of better as long as we only have the competing GPU designers pushing the state of the art.

 

[3DVagabond]

Show me a console game with better I.Q than a PC game.
It's that simple.

Do you think this is because the rendering API is superior on PC? Because that;s what this thread is about. It's not PC vs. Console.

 

[3DVagabond]

You don know either that AMD's DX11 dosn't support multithreading and never has?
Is this really forgotten?

http://forums.anandtech.com/showpost.php?p=31520674&postcount=28

I think you should read a bit up (using that link) on what multithreading in the DX11 drivers does....Try finding the phrase: "Traditionally, rendering is a very serial process..."

Mantle and DX11 multithreading sounds to offer some similar "benefits"...one can wonder why AMD dosn't support the function...for EVEN mor added performance?

Does it look like I am kidding now?
Can you find more vaild arguments?

This is not about DX11 multi threading support. Please try and comment on Mantle. This is the only Mantle thread we are allowed. It would be nice to not have to wade through posts about nVidia features all of the time.

 

[nenforcer]

This is not about DX11 multi threading support. Please try and comment on Mantle. This is the only Mantle thread we are allowed. It would be nice to not have to wade through posts about nVidia features all of the time.

Exactly, I don't understand why people are so upset that AMD doesn't have a multithreaded DX11 driver (YET!). If the supposed performance gain is only up to 2X the number of draw calls then clearly they designed Mantle to eliminate what is the true bottleneck, DirectX.

AMD is claiming up to 10X the number of draw calls than DX11 in games that support Mantle. I'm waiting for these new BF4 Mantle benchmarks before I purchase a new video card.

I'm glad AMD is finally doing things to differentiate their graphics cards (Mantle, TrueAudio) in order to compete with nVidia's own proprietary libraries (PhysX, CUDA).

 

[Red Hawk]

If Mantle truly provides the performance advantages AMD has talked up, then I'm glad if they diverted resources away from enabling DirectX 11 multithreaded rendering over to Mantle.

 

[krumme]

I'm sure when they were asking for it they were hoping it would work on more than the 5% of the market GCN represents.

Trash talk and trolling. Go away bella if you dont want to contribute.