AMD's new Mantle API

[erocker]

Well in my opinion if Mantle will become widely accepted among game devs in upcomming months/ years I don't see why AMD wouldnt share it with Nvidia for let's say PhysX / CUDA support on AMD's gpus. What do you think guys?

I think a low-level API is something Nvidia can do on their own. They don't need any help from AMD.

 

[Frag_Maniac]

I think a low-level API is something Nvidia can do on their own. They don't need any help from AMD.

The reasons they haven't are...

1. Their answer is to offer uber expensive cards to get better performance

2. They have no console hardware that games are made for then ported to PC

Now they will have a reason to though if Mantle takes off. Nvidia have been flying high by fostering the illusion that games can only be so optimized, requiring customers to put up the cash and buy big hardware. Mantle changes that way of thinking (or should I say brainwash?). It's easy to see why Nvidia have not been interested in making anything like Mantle. They're sitting pretty on inflated profits from enthusiast gamers whom insist their hardware is better.

 

[ruff0r]

i like where this is going - i just remember the days when glide sucked for having the 'wrong' graphics card

Exactly my thoughts. What if Nvidia is not jumping onto the "Mantle" Boat? But everyone else does will our DirectX cards suffer huge from that?

other thoughts i hope it will be free accessible for coders...maybe Linux support? new platform for gamers? Linux+Mantle?

 

[Fourstaff]

One more piece of code to write up, I don't think many will bother unless AMD provides coders.

 

[TRWOV]

I don't think developers will have to code themselves, rather there will be a Mantle compiler built in the engine.

 

[ST.Viper]

a low-level API is something Nvidia can do on their own

Hope that never happen. Nothing against Nvidia, but imagine that in that case we will have amd / nvidia exclusive game titles or engines as they will not support both gpu vendors unless game developers will bother to code pc game two times and than port game to the consoles which is unlikely. So in the end they will just stick with DirectX. So for realy good looking and smooth playing game you will need a powerfull graphics card or two (again ).

 

[FrankWest]

I'll just leave this here.

 

[W1zzard]

I'll just leave this here.

http://img.expreview.com/news/2013/09/29/mantle.jpg

you don't have to do anything. if your app is not cpu limited there is nothing to gain.

of course you could argue that mantle = less cpu overhead = more cpu time available for fx = better graphics (even more work)

unless amd brings me a suitcase full of money i'd rather have my coders work on more game relevant stuff

 

[Aquinus]

you don't have to do anything. if your app is not cpu limited there is nothing to gain.

I think that some people would be pretty excited to hear that they don't need to update their aging CPU as well as their GPU just to play the latest games really well. Also, why get a super beefy Intel i7 when an APU could possibly run a game just as well with the same GPU?

I think an argument to be made here is that AMD is trying to show that they're making gaming cheaper; trying to show that you don't need a super expensive gaming rig to play games with nice graphics.

 

[mastrdrver]

you don't have to do anything. if your app is not cpu limited there is nothing to gain.

of course you could argue that mantle = less cpu overhead = more cpu time available for fx = better graphics (even more work)

unless amd brings me a suitcase full of money i'd rather have my coders work on more game relevant stuff

I think this has more to do with hUMA when asking why AMD is doing this API. The only reason it will have any success is because AMD has both the CPU and GPU in the PS4 and XB1. Not saying that PS4 and/or XB2 will be using hUMA, but an AMD employee linked the Anand article (about Mantle) in his twitter saying they were on the right track.

 

[W1zzard]

I think that some people would be pretty excited to hear that they don't need to update their aging CPU as well as their GPU just to play the latest games really well. Also, why get a super beefy Intel i7 when an APU could possibly run a game just as well with the same GPU?

as the name suggests hardware companies are in the business of selling hardware, so this will never happen

 

[bencrutz]

unless amd brings me a suitcase full of money i'd rather have my coders work on more game relevant stuff

well, you're not a developer working on titles for ps4 / xbox one
coz if you are, i think youd be grateful to have mantle on PC, ofc assuming mantle is really working as it supposed to

 

[crazyeyesreaper]

and you are talking to a man that codes more shit than probably anyone on this site lulz I would expect W1zz to have idea of what to expect. Just as actually game developers / designs do and while Mantle is a great idea I fail to see it taking off. After all its just a continuation of AMDs older Stream and CTM that never took off in the first place.

THe problem isnt the idea it is usually AMDs execution. Regardless while the Mantle API could be useful without backing from Nvidia it wont be rapidly adopted and as such will eventually fail to gain traction.

To put it bluntly Nvidia still controls greater than 50% at 52% of the GPU market according to the Steam Hardware survey. While AMD sits at just 33% and Intel at 14%

Considering their competitiors hardware = 66% of the steam user base or 2 / 3 AMD has alot they need to do before they can truly push Mantle. Developers cannot and will not risk PC sales by using Mantle in serious ways. It will be more like Nvidia's PhysX in that it adds more pretty or performance is better but ends up being completely uneccessary to run and enjoy the game.

 

[Frick]

THe problem isnt the idea it is usually AMDs execution. Regardless while the Mantle API could be useful without backing from Nvidia it wont be rapidly adopted and as such will eventually fail to gain traction.

To put it bluntly Nvidia still controls greater than 50% at 52% of the GPU market according to the Steam Hardware survey. While AMD sits at just 33% and Intel at 14%

I think you're forgetting the consoles here. They are what makes it interesting.

 

[Frag_Maniac]

What I want to see is a reliable 3rd party source test their claim of Mantle's 9x the draw calls, because if that's true, it's certainly not as trivial as some are making it out to be, regardless of the learning cure. Yeah it may mean a bit more production time until they learn how to use it, but the long term benefits would far outweigh the time spent getting used to it.

I also don't get why so many are worried about going away from DirectX. Has DirectX 11 really been implemented to it's full potential in most games out there? Is it any better really? Look at all the performance problems in many of the games that use it, and tell me in all honesty you'd really miss it.

As far as I'm concerned Mantle is a bit like SteamOS, an alternative to Microsoft's way. A bold step, but could potentially be a very good one.

 

[Recus]

My thoughts:. Power boost for next gen consoles delayed (#1, #2, #3, #4, #5) so AMD wants better performance for PC. Does it matter since most monitors are 60 refresh rate?

 

[crazyeyesreaper]

I think you're forgetting the consoles here. They are what makes it interesting.

I never said it wasn't interesting I just don't see developers taking it all that seriously maybe 3-4 years from now but even then that will be due to all new game engines from the ground up.

Key things to look at it

It requires game engine integration which takes time and money not only that you then need to educate your coders and team i general on how to make the best use of the API documentation for this also takes time. Building tools to take advantage of it takes time.

The API needs proper support from the ground up. Considering AMD has a hard time even getting their multi-gpu configurations working properly you can say I will be mighty impressed if this ends up as something more than a bullet point on the back of the GPU box.

Also while it takes advantage of GCN should AMD abandon GCN in the future and change their design philosophy then what? Direct X and Open GL may not be as efficient but it cares less about the architecture. So while AMD does have dominance in the console market their ability to force that on the PC market comes down to execution and market share which are two things AMD lacks.

 

[Frick]

I never said it wasn't interesting I just don't see developers taking it all that seriously maybe 3-4 years from now but even then that will be due to all new game engines from the ground up.

Key things to look at it

It requires game engine integration which takes time and money not only that you then need to educate your coders and team i general on how to make the best use of the API documentation for this also takes time. Building tools to take advantage of it takes time.

The API needs proper support from the ground up. Considering AMD has a hard time even getting their multi-gpu configurations working properly you can say I will be mighty impressed if this ends up as something more than a bullet point on the back of the GPU box.

Also while it takes advantage of GCN should AMD abandon GCN in the future and change their design philosophy then what? Direct X and Open GL may not be as efficient but it cares less about the architecture. So while AMD does have dominance in the console market their ability to force that on the PC market comes down to execution and market share which are two things AMD lacks.

All good points.

Can't wait for mid 2014, I want to see how this pans out!

 

[crazyeyesreaper]

Your not alone Frick I myself am also very interested in how this turns out but having an education centered around game design and art creation. Its gonna take alot more than fancy GPUs in the consoles and slick promises to get Developers on board. However if AMD does succeed I would be impressed and rather happy as it can only do good things for their bottom line and help keep them afloat.

 

[TRWOV]

I think that AMD blew the whistle too soon. They should have shipped the games and let the press find out later why FB3 based games perform way better on GCN. That would a) catch nVidia off-guard and b) let the results speak for themselves.

The only thing they've left with this announcement are lots of unknowns. Some developers (like the tweet Frag Maniac quoted) think that they'll have to learn new code which, if true, would pretty much spell doom for the initiative. This will force AMD to spill out more info than necessary in order to dismantle rumors.

 

[TheoneandonlyMrK]

I never said it wasn't interesting I just don't see developers taking it all that seriously maybe 3-4 years from now but even then that will be due to all new game engines from the ground up.

Key things to look at it

It requires game engine integration which takes time and money not only that you then need to educate your coders and team i general on how to make the best use of the API documentation for this also takes time. Building tools to take advantage of it takes time.

The API needs proper support from the ground up. Considering AMD has a hard time even getting their multi-gpu configurations working properly you can say I will be mighty impressed if this ends up as something more than a bullet point on the back of the GPU box.

Also while it takes advantage of GCN should AMD abandon GCN in the future and change their design philosophy then what? Direct X and Open GL may not be as efficient but it cares less about the architecture. So while AMD does have dominance in the console market their ability to force that on the PC market comes down to execution and market share which are two things AMD lacks.

I think your putting too much on Amd's shoulders. They will get most of the drive they need and more than enough traction regardless due to the same Api being identical across 4 formats 3 being console ,


Thing is, if Bf4 stomps ass with the Api, or even a later game really shows its worth some other devs are then getting pushed into using it to regain an edge

I do get your point that no one is forced to use it but all EA games are now likely to use it including starwars ,crytek are in and the unreal engines onboard so the devs will eventually get it for free to use or not and you all thought it pondersus that nvidia has got all cosy with linus ,,pretty soon dx is a dodo opengl lives on imho.

 

[Xzibit]

AMD having "won" console contracts is nothing new, and it's more due to Nvidia being unwilling to compete in that arena than AMD "winning" it from anyone. Indeed, Nvidia have made it easy for AMD to get exclusive console contracts year after year by their apathy toward that market and it's lesser profit margins, and it may come back to bite them in the ass eventually.

I think this explains it better than "Nvidia is unwilling to compete"

[Forbes]: The Real Reasons Microsoft, Sony Chose AMD For The Xbox One And PS4

My sources have confirmed for me that both Sony and Microsoft felt that MIPS didn’t have the right size developer ecosystem or the horsepower to power the new consoles.

At the end of the bake-off, ARM was deemed as not having the right kind of horsepower and that its 64-bit architecture wasn’t ready soon enough.

The requirement for an X86-based SOC ostensibly removed Nvidia from the running.

The requirement for a custom SOC removed Intel from the running, as well as their graphics.

It was a clear-cut win.

How could Nvidia even compete. If they ever had a chance it was with a separate GPU and that was a no-no.
The closes thing they had was Wayne Tegra 4 and that's barely doing 20FPS @ 1080P. Which Qualcomm Snapdragon 800 outperforms it.

It was handed to AMD since they were in the best position and its up to them to take advantage or exploit it to their benefit.

They mentioned Mantle is cross platform so it can be X-Bone - PS4 - Windows and maybe Linux - Nintendo - iOS - Android.

They can either make it work or screw it up. Time will tell.

 

[crazyeyesreaper]

I think your putting too much on Amd's shoulders. They will get most of the drive they need and more than enough traction regardless due to the same Api being identical across 4 formats 3 being console ,


Thing is, if Bf4 stomps ass with the Api, or even a later game really shows its worth some other devs are then getting pushed into using it to regain an edge

I do get your point that no one is forced to use it but all EA games are now likely to use it including starwars ,crytek are in and the unreal engines onboard so the devs will eventually get it for free to use or not and you all thought it pondersus that nvidia has got all cosy with linus ,,pretty soon dx is a dodo opengl lives on imho.

And yet if this API can't deliver consistant massive performance gains over Direct X then whats the point. All we currently have are vague promises of performance increases on GCN

but where is GCN going to be 5 years from now?

Technically this tech already exists and AMD failed to capitlize on it due to Microsoft changing how Windows worked when it comes to AMD Stream changes in Windows 8 opened this avenues back up.

I just don't see it taking off.

So what if Crytek is onboard most of their games are generic crap now and they tend to have a disgain for PC gamers and are planning to go F2P with all titles this means limiting graphical advancements in order to garner the largest user base so Crytek is not a major player in this.

Unreal Engine hurray the most widely used engine where all games look pretty much the same with overly steroid monster male protagonists. In the PC space Unreal Engine games that work well and without issue is rare especially with the crappy mouse auto smoothing that some games dont even allow you to turn off that makes controls seem sluggish and unresponsive. Granted with UE support thats a feather in AMDs cap but doesnt really impress me that much since UE hasnt changed much at all since version 2.0

Dice and Frostbite is a rather large aquisition for Mantle but considering Bioware has consistantly been alienating their fan base / DICE is moving ever closer to being a CoD clone and most of EA's racing games being crap I don't see much here to get excited about other than the fact the Frostbite engine itself is rather amazing.

EA and their in house studios Crytek and their free to play and crappy UE engine ports to PC will not make Mantle shine.

Battlefield 4 will be a good booster shot but they need more than that. They need new franchises and fresh IPs games that are actually good not just inflated review scores.

In a world where this is common
http://gamasutra.com/view/news/1659...t_New_Vegas_bonus_by_one_Metacritic_point.php

http://gamasutra.com/view/feature/169354/how_creative_assemblys_process_.php

I don't have much hope in Mantle I love great graphics but I find myself going back in time to play truly great games as most of todays modern titles are just fucking garbage.

If the Mantle API can boost performance by more than 25% and comes with fantastic games thats great and I will be impressed but mainstream regurgitated garbage with a new coat of paint doesn't impress me.

Lets see if

CD Project Red
Bethesda
Obsidian
Creative Assembly
Valve
Blizzard
etc pick up Mantle

Just because EA and their acquired studios may use mantle doesnt mean it will be successful. I found the games from above to be far more fun than most of what gets pushed out today. But to each their own.

Mantle is a very interesting proposition I just don't see it being successful long term. It has alot of the same traits to and inherent problems that PhysX had and still has it can run on this or that but excluding or having key players avoid it = creates problem and slows adoption and usefulness.

If Nvidia doesnt support Mantle and AMD eventually gets a graphics monopoly then what happens to GPU prices and innovation? Theres alot going on here both on the surface and behind the scene i can't wait to see it unfold but I am a skeptic as this isnt the first time AMD has tried to push something and failed miserably at it.

Do I need to remind people that Tessellation on AMD hardware has existed for a very very long time but never caught on due to functionality?
Or that Mantle is built off the failed CTM / Stream technologies that AMD failed to effectively market to begin with and then got cut off by Microsoft only for them to change their minds again? Rose-tinted glasses make the world a better place I suppose but I am a realest and AMD has a LONG way to go with this.

 

[The Von Matrices]

On a purely scientific note, I take issue with them calling this "mantle." Going with the metaphor, they're only eliminating the crust, which is at most is 0.8% of the earth's diameter. So does this mean that we should only expect a 0.8% reduction in overhead? If they had called it "core" then although it would be generic it would make more sense.

 

[TheoneandonlyMrK]

And yet if this API can't deliver consistant massive performance gains over Direct X then whats the point. All we currently have are vague promises of performance increases on GCN

but where is GCN going to be 5 years from now?

Technically this tech already exists and AMD failed to capitlize on it due to Microsoft changing how Windows worked when it comes to AMD Stream changes in Windows 8 opened this avenues back up.

I just don't see it taking off.

So what if Crytek is onboard most of their games are generic crap now and they tend to have a disgain for PC gamers and are planning to go F2P with all titles this means limiting graphical advancements in order to garner the largest user base so Crytek is not a major player in this.

Unreal Engine hurray the most widely used engine where all games look pretty much the same with overly steroid monster male protagonists. In the PC space Unreal Engine games that work well and without issue is rare especially with the crappy mouse auto smoothing that some games dont even allow you to turn off that makes controls seem sluggish and unresponsive. Granted with UE support thats a feather in AMDs cap but doesnt really impress me that much since UE hasnt changed much at all since version 2.0

Dice and Frostbite is a rather large aquisition for Mantle but considering Bioware has consistantly been alienating their fan base / DICE is moving ever closer to being a CoD clone and most of EA's racing games being crap I don't see much here to get excited about other than the fact the Frostbite engine itself is rather amazing.

EA and their in house studios Crytek and their free to play and crappy UE engine ports to PC will not make Mantle shine.

Battlefield 4 will be a good booster shot but they need more than that. They need new franchises and fresh IPs games that are actually good not just inflated review scores.

In a world where this is common
http://gamasutra.com/view/news/1659...t_New_Vegas_bonus_by_one_Metacritic_point.php

http://gamasutra.com/view/feature/169354/how_creative_assemblys_process_.php

I don't have much hope in Mantle I love great graphics but I find myself going back in time to play truly great games as most of todays modern titles are just fucking garbage.

If the Mantle API can boost performance by more than 25% and comes with fantastic games thats great and I will be impressed but mainstream regurgitated garbage with a new coat of paint doesn't impress me.

Lets see if

CD Project Red
Bethesda
Obsidian
Creative Assembly
Valve
Blizzard
etc pick up Mantle

Just because EA and their acquired studios may use mantle doesnt mean it will be successful. I found the games from above to be far more fun than most of what gets pushed out today. But to each their own.

Mantle is a very interesting proposition I just don't see it being successful long term. It has alot of the same traits to and inherent problems that PhysX had and still has it can run on this or that but excluding or having key players avoid it creates problems

If Nvidia doesnt support Mantle and AMD eventually gets a graphics monopoly then what happens to GPU prices and innovation? Theres alot going on here both on the surface and behind the scene i can't wait to see it unfold but I am a skeptic as this isnt the first time AMD has tried to push something and failed miserably at it.

Do I need to remind people that Tessellation on AMD hardware has existed for a very very long time but never caught on due to functionality?
Or that Mantle is built off the failed CTM / Stream technologies that AMD failed to effectively market to begin with and then got cut off by Microsoft only for them to change their minds again? Rose-tinted glasses make the world a better place I suppose but I am a realest and AMD has a LONG way to go with this.

Amd often does things first and you might ask yourself how long it would have taken for tesselation to take off had nvidia got there first,,, you seen a lot of physx games i can run on my hybrid system no, neither have i , and how did stream fail its been an evolving process mantle has little to do with anyway but stream was a start into gpgpu land.

Mantel is not stream, it is not a language wherein gpu compute can be done, its a driver ONE driver to rule them all

dont get beat up about intel and nvidia ,they have oportunities in this too , just not easy to swallow ones.

put it this way with ten year console life spans im expecting a massive group hug betwixt nvidia ,intel and amd in about a year or 3