AMD Mantle

[Stringjam]

I now have to incorporate an extra effort in developing to ensure a less than majority of gamers get more performance from my game than those that use strictly DirectX. More time developing, more complexity, more versions, more money spent with no real return on investment.

Big distinction if I'm a developer.

If Mantle is the API that AMD is using on consoles, then wouldn't it be safe to assume that nearly every developer will already be gearing up for it?

The article speculated at this possibility, that if this were actually the case, much of the work spent developing the game for the console would directly translate to the PC version.

 

[Markbnj]

I'm not honestly sure why this matters anymore, when PCs are already better than the current generation consoles that games are actually developed for (before being sloppily ported to the PC to pick up a couple extra bucks). It's not like there is a long line of devs out there just waiting for another 30% performance gain on the PC platform so they can blow us away.

 

[CuriousMike]

I now have to incorporate an extra effort in developing to ensure a less than majority of gamers get more performance from my game than those that use strictly DirectX. More time developing, more complexity, more versions, more money spent with no real return on investment.

Big distinction if I'm a developer.

You're assuming the development team won't settle on Mantle as their API at the beginning.

If they do, then the benefits are free for all platforms, not just the lead SKU.

 

[blackened23]

I think I enjoy the Mantle discussion here more than the one in the GPU sub-forum. Tons of people getting mad and arguing over video cards over there...christ...it's almost like a religion to those guys on both sides (nvidia and AMD fans).

 

[blackened23]

I now have to incorporate an extra effort in developing to ensure a less than majority of gamers get more performance from my game than those that use strictly DirectX. More time developing, more complexity, more versions, more money spent with no real return on investment.

Big distinction if I'm a developer.

Frostbite 3 will have Mantle baked in for all developers to use from the get-go. Other engines apparently are getting similar functionality if AMD is to be believed.

It should be mentioned that EA is leveraging the Frostbite 3 engine to be used in quite a few of its games - including Dragon Age 3, the next Star Wars game, among many others.

The other issue is that the rumor is - Mantle is very similar to the APIs used on the next generation consoles. If that is the case, that could very well increase the adoption rate by quite a bit - and that would separate it from "glide status". If anyone here remembers, Glide was created circa 1996 when there were zero standards. There was the rendition verite, there was 3dfx, and there was powerVR and THEY ALL had different rendering APIs. I'm sure i'm missing a few other APIs as well - Direct3D made sense then because there were 6 competing APIs which had nothing to do with the then-released consoles.

As compared to now? There is one API, D3D, and there are the consoles. Presumably, the next-gen consoles will be using APIs similar to mantle since AMD has "control" over the hardware, so to speak. That is the critical difference here and where Mantle distinguishes itself from glide - these are different times, there are not 6 different APIs competing with each other; there is just one. If Mantle is indeed being used in the next-generation consoles (either of them) it will be huge for AMD. This is aside from the fact that Frostbite 3 will have Mantle baked in for all developers to use - and EA is using frostbite 3 in quite a few titles, not just BF4.

I suppose we'll have to wait and see how the mantle version of BF4 performs, but this has potential for AMD.

 

[Aikouka]

The other issue is that the rumor is - Mantle is very similar to the APIs used on the next generation consoles.

This has always come across as strange to me because it always sounded like Sony and Microsoft wrote their own low-level drivers for their respective consoles. Now, they most likely had some assistance from AMD given AMD's architecture knowledge would be invaluable, but I'm kind of doubting that there will be a 1:1 code mapping like people seem to be suggesting.

That is the critical difference here and where Mantle distinguishes itself from glide - these are different times, there are not 6 different APIs competing with each other; there is just one.

Mantle is also adaptable with any card using HLSL. Although, I can't see NVIDIA spending time to get it working on Kepler/Fermi when they're planning on releasing Maxwell (quite the different architecture) next year.

 

[colonelciller]

The article mentions that Frostbite 3 is going to be capable of rendering natively to Mantle.

Sooooo......one of the biggest game development engines has made an effort to accommodate a render path that totally excludes the need for Direct3D abstraction?

The SteamBox is suddenly looking a whole lot more viable in the future if this becomes the case.

What if Unreal and CryEngine decide to follow suit? Suddenly we're seeing AAA titles that no longer need Windows.

ohhh... this is getting good!
sayonara windows

 

[colonelciller]

1. Frostbite 3 uses Mantle for amd cards. But its not exclusively all mantle.
2. It uses DX. It has to since it needs to support other cards.
3. steambox does not have anything to do with it. That is its own thing, might as well call it OnLive, in the house.
4. Again, those games will still have to use DX for games.

i think you're missing a critical point.
if this enables devs to develop for consoles AND linux with little to no additional effort, them it matters zilch if they are still developing for DX...

...because people will begin to migrate to linux &/or steam box for gaming.
not everyone but a sizeable number may...putting further pressure on devs to develop for linux.. creating more freedom in the OS space for people to migrate to linux.

now the sh!tty UEFI locks on some new computers are showing its true intended purpose... to keep people on windows

 

[Haider]

It's the low-level API which is what AMD probably provided to Sony and MS. Since the AMD PC cards use the same architecture. It should also be relatively easy to transport the console code across. Used to transport 68000 assembler done on an Amiga to a Mac at Uni. Finally this may resolve the draw calls issue on PCs and we can get GTA series to run well

 

[smackababy]

From what I'm understand, even with low level API access, in order to do anything, you still need to perform processing to the data you're sending, thus you either use a high level API or write your own.

If Mantle is just a low level API, I can't see many studios using it, due to the fact they would have to use a high level API like DX for non AMD cards anyway. Why spend the extra resources so AMD gets better performance and possibly segment over half of their prospective market.

Also, Mantle is probably similar to the console API, but I doubt it is the same. MS and Sony likely created theirs along side AMD, who then used that to create Mantle.

 

[Stringjam]

If Mantle is just a low level API, I can't see many studios using it, due to the fact they would have to use a high level API like DX for non AMD cards anyway. Why spend the extra resources so AMD gets better performance and possibly segment over half of their prospective market.

I might agree, if not for some very big players who are taking it quite seriously, such as DICE / EA specifically working with AMD to implement it in Frostbite3 (which will make its debut in December in a BF4 update).

There are also a few comments floating around from Crytek employees that it is something they might consider (in typical, vague, Crytek fashion).

I'm sure we'll be quite a bit more aware of the scope of this thing by the end of the year. Whole lot of speculation floating about right now.

 

[smackababy]

I might agree, if not for some very big players who are taking it quite seriously, such as DICE / EA specifically working with AMD to implement it in Frostbite3 (which will make its debut in December in a BF4 update).

There are also a few comments floating around from Crytek employees that it is something they might consider (in typical, vague, Crytek fashion).

I'm sure we'll be quite a bit more aware of the scope of this thing by the end of the year. Whole lot of speculation floating about right now.

The problem is developers don't care what GPU you have. Why would they put a lot more resources into making a game play better on a certain (minority) graphics card? They will still have to release a DX version, so why wouldn't they just put all the resources they could have spent on Mantle optimizations, to make the DX version that much better?

I can see AMD giving some developers money and helping implement this, but unless the DX version is pretty close, the game will flop. Look at Steam stats; the majority of users are Nvidia users.

If AMD wants this to take off, they need to give developers incentive to use Mantle, and then in a few years (after a ton of games utilize it), it could factor into the average consumer's decision of which card to buy. And that is also dependent on the real performance increase we see. If developers have to write their own high level implementation, it will have the potential to be better than DX, but also the potential to be worse.

 

[Staples]

I think you guys are making too big a deal about Frostbite using Mantle.

Which games besides the Battlefield series even used Frostbite? None that I know of and it isn't an engine that can be licensed to other developers.

I'd be much more excited if Unreal Engine supported it because the Unreal Engine is very widely used on all sorts of games.

Anyway, I am sure Unreal will add it to their engine soon because it would be a huge selling point to potential licensees.

 

[smackababy]

Battlefront 3 is going to use Frostbite. But even then, the Mantle stuff isn't baked into the engine. They are patching it in.

 

[gorobei]

I think you guys are making too big a deal about Frostbite using Mantle.

Which games besides the Battlefield series even used Frostbite? None that I know of and it isn't an engine that can be licensed to other developers.

umm you might want to do some research before you talk.
http://en.wikipedia.org/wiki/Frostbite_(game_engine)

Frostbite 3 is the next generation version of Frostbite and is powering the upcoming games Battlefield 4 (set to release October 29th of 2013), Need for Speed: Rivals (to be released in November, 2013), Star Wars: Battlefront (set to release the Q2/Q3 of 2015) and the prequel to Mirror's Edge. The game engine has several upgrades including improved tessellation technology. It also features Destruction 4.0, which enhances the in-game destruction over its predecessors.[16] Frostbite 3 will also power Dragon Age: Inquisition, Plants vs. Zombies: Garden Warfare and the next Mass Effect.

its kinda common knowledge that EA is using frostbite for all of their in house titles.

 

[imaheadcase]

i think you're missing a critical point.
if this enables devs to develop for consoles AND linux with little to no additional effort, them it matters zilch if they are still developing for DX...

...because people will begin to migrate to linux &/or steam box for gaming.
not everyone but a sizeable number may...putting further pressure on devs to develop for linux.. creating more freedom in the OS space for people to migrate to linux.

now the sh!tty UEFI locks on some new computers are showing its true intended purpose... to keep people on windows

Mmm no i'm not missing the point. The point is its going to take just more than Mantle to make devs care about linux. Its not drivers that make people not game on linux, its linux.

 

[Imaginer]

Seems to me the reason why people care about Linux and the reinvention of one more standard to add to developing for, is that for some reason, people have a bias disdain for Windows for gaming.

Not to say a lack of additional options is a good thing, but I feel some are solely joining this reinventing the wheel and API is because of this reason.

I personally would rather continue to use something that works and continues to work, spending more effort in the forward progress already laid out than to go back and do it all over again (akin back to when the fight between Glide and wrappers were done in the Pentium era)

 

[smackababy]

Seems to me the reason why people care about Linux and the reinvention of one more standard to add to developing for, is that for some reason, people have a bias disdain for Windows for gaming.

Not to say a lack of additional options is a good thing, but I feel some are solely joining this reinventing the wheel and API is because of this reason.

I personally would rather continue to use something that works and continues to work, spending more effort in the forward progress already laid out than to go back and do it all over again (akin back to when the fight between Glide and wrappers were done in the Pentium era)

A lot of the people talking up Mantle are AMD fanoboys. The threads in VC&G are pretty moronic. They seem to believe a mid range AMD card will now double Titan performance with Mantle, or that somehow 4k gaming on Ultra with a 7970 is possible.

The Windows hate is mostly unfounded, but some of the concern isn't entirely off base. With Windows 8's App Store thing, there is a semi-valid concern MS might decided to stop letting 'side loading' of programs. Realistically, this makes very little sense from an enterprise aspect (where Windows makes the majority of it's money). The majority of people who hate Windows, hate it because Gabe said "Windows 8 sux!" and so they just jump on board, or they dislike the UI direction MS is heading.

The move for Valve makes sense because eventually MS will be able to compete with them from the digital store front aspect. Valve needs a place it can 'flourish' and try to lure gamers to. If Valve gets big on Linux and can switch gamers to it, Windows can't compete with a Microsoft Linux store, that just don't make any sense. Valve can keep the servers on and collect the billions.

 

[Imaginer]

The move for Valve makes sense because eventually MS will be able to compete with them from the digital store front aspect. Valve needs a place it can 'flourish' and try to lure gamers to. If Valve gets big on Linux and can switch gamers to it, Windows can't compete with a Microsoft Linux store, that just don't make any sense. Valve can keep the servers on and collect the billions.

Here is the other kicker, Microsoft's last hurrah with PC gaming came with Games for Windows, and a split between that direction and the Xbox. Steam should not feel threatened as they already established themselves as the de-facto standard of Windows PC gaming - consolidating it before Microsoft had an official Windows app front (note I said app, NOT applications).


So yes, Gabe is in my opinion, in part throwing tantrum, and giving Valve more options, much like Microsoft did with Windows RT.

And there SHOULD not be any fear of not being able to install applications not officially on the store (almost and all desktop applications are in this category - unless RT and all)

The real reason to have the disdain, is the RT version of Windows where those fears DO stand. But Microsoft is in that very position Steam is in, but with hardware instead of software and from the likes of Apple and Android competitors.

A very very complicated dance indeed.

 

[Staples]

umm you might want to do some research before you talk.
http://en.wikipedia.org/wiki/Frostbite_(game_engine)


its kinda common knowledge that EA is using frostbite for all of their in house titles.

Thank you for correcting me. I didn't know this but I have forever believed that publishers need to flex their muscle with game engines they own. Glad to see that it is finally becoming a reality.

 

[Stringjam]

I personally would rather continue to use something that works and continues to work, spending more effort in the forward progress already laid out than to go back and do it all over again (akin back to when the fight between Glide and wrappers were done in the Pentium era)

How well is it really working, though? I can get into CryEngine3 and put together a level and my framerate drops like a rock when I get above 3000 draw calls, which is child's play for a PlayStation. Despite the fact that my hardware is light years ahead of what's in that box.

Mantle is claiming up to a 9x increase in the DP allowance.....that is HUGE. I don't even own an AMD product so I'm not a "fan-boy," but if this ended up actually being true, it would be a leap forward for PC performance and how much detail the environment artists can work with.

 

[frozentundra123456]

i think you're missing a critical point.
if this enables devs to develop for consoles AND linux with little to no additional effort, them it matters zilch if they are still developing for DX...

...because people will begin to migrate to linux &/or steam box for gaming.
not everyone but a sizeable number may...putting further pressure on devs to develop for linux.. creating more freedom in the OS space for people to migrate to linux.

now the sh!tty UEFI locks on some new computers are showing its true intended purpose... to keep people on windows

Even if Mantle takes off, it would only apply to new games. There will still be a lot (the vast majority actually) of games that will require windows (DX). I am sure there will also probably be new PC exclusives and a lot of indie games that will also not utilize Mantle. So unless you want to limit yourself to a rather small subset of games (i.e. new games coming out that support Mantle) I would not so gleefully dismiss windows.

 

[smackababy]

How well is it really working, though? I can get into CryEngine3 and put together a level and my framerate drops like a rock when I get above 3000 draw calls, which is child's play for a PlayStation. Despite the fact that my hardware is light years ahead of what's in that box.

Mantle is claiming up to a 9x increase in the DP allowance.....that is HUGE. I don't even own an AMD product so I'm not a "fan-boy," but if this ended up actually being true, it would be a leap forward for PC performance and how much detail the environment artists can work with.

Do we really know what they are doing with those calls though? If they are doing 9x that with the same amount of processing required in your CryEngine level and keeping the same frame rates, yes that is impressive. However, like almost all benchmark claims by the SELLER of the product, they are under extremely narrow scenarios and very deceptive.

 

[Imaginer]

How well is it really working, though? I can get into CryEngine3 and put together a level and my framerate drops like a rock when I get above 3000 draw calls, which is child's play for a PlayStation. Despite the fact that my hardware is light years ahead of what's in that box.

Mantle is claiming up to a 9x increase in the DP allowance.....that is HUGE. I don't even own an AMD product so I'm not a "fan-boy," but if this ended up actually being true, it would be a leap forward for PC performance and how much detail the environment artists can work with.

What about Intel derivative GPU solutions? Such as in a mobile solution I have adapted. Sorry, but if it does not work across the board, it can't really that too exciting in my mind.

And I have Two 7970s. They work fine as is. My HTPC has a 6950, works fine as is.

 

[james1701]

I would really like to hear from Crytech and see if they are going to incorporate Mantle.