The AMD Mantle Thread

[ams23]

Unless proven otherwise, the fact remains that Mantle is, for the time being, IHV-specific, GCN-specific, requires more software development work compared to using industry standard graphics API's, and is not officially endorsed nor supported by Sony and Microsoft in next gen consoles. As industry standard graphics API's evolve and improve to take better advantage of upcoming architectures where CPU and GPU are more closely integrated, Mantle will become increasingly less relevant.

 

[Despoiler]

the fact remains that Mantle is, for the time being, IHV-specific, GCN-specific,

True

requires more software development work compared to using industry standard graphics API's,

Mostly not true. Anything developed for console to be ported to PC or developed for PC to be ported to consoles will have savings in development time. Only games that are not being ported will have more development time.

and is not officially endorsed nor supported by Sony and Microsoft in next gen consoles.

How does this matter? The consoles have low level access to the GPU. They don't need Mantle. Mantle is giving PCs low level access like consoles have.

As industry standard graphics API's evolve and improve to take better advantage of upcoming architectures where CPU and GPU are more closely integrated, Mantle will become increasingly less relevant.

I would say this wildly inaccurate in the case of DirectX as it has become more and more bloated and increase overhead with each successive version. MS has proven time and time again that they just don't get anything, but whatever they think something should be. That thought misses the mark time and time again. I have no hope that Mantle forces them to react to clean up DirectX. OpenGL on the other hand will eventually have Mantle commands natively supported according to AMD. At least that is their goal.

 

[ams23]

Not true. Anything developed for ported to PC or developed to PC t consoles will have savings in development time

Incorrect. Games ported to and from consoles will rely on higher level industry standard API's to reach as wide an audience as possible. Software development work for Mantle will be extra and will only target a very small proportion of PC gaming systems (especially in the near future).

 

[desprado]

Incorrect. Games ported to and from consoles will rely on higher level industry standard API's to reach as wide an audience as possible. Software development work for Mantle will be extra and will only target a very small proportion of PC gaming systems (especially in the near future).

A simple fact which some people having difficulty to understand.

 

[ams23]

On a side note, AMD has perhaps (unknowingly?) dealt a blow to the future of game consoles. By commoditizing major features of game consoles (such as lower level API access, Audio processing, etc.) to put on a PC , there is less of a need to use a console in the first place (short of console exclusive games). A significant market still exists for consoles, but it is shrinking.

 

[SiliconWars]

A simple fact which some people having difficulty to understand.

The reason people are having difficulty understanding this "simple fact" is that more devs keep signing up for Mantle then wax lyrically about how awesome it is.

It's now been added to what, 4 or 5 game engines? Considering it's not even in beta yet that's unreal. Pun intended.

 

[Paul98]

Incorrect. Games ported to and from consoles will rely on higher level industry standard API's to reach as wide an audience as possible. Software development work for Mantle will be extra and will only target a very small proportion of PC gaming systems (especially in the near future).

The point has been that Mantle porting from console will take very little development time. It will give better performance and stability than the DX version.

 

[SiliconWars]

Incorrect. Games ported to and from consoles will rely on higher level industry standard API's to reach as wide an audience as possible. Software development work for Mantle will be extra and will only target a very small proportion of PC gaming systems (especially in the near future).

Or maybe the devs will just do less optimizing on the old and broken DX while doing more on Mantle instead.

 

[Paul98]

How much would mantle speed up porting games if developers didn't have to go DX for the rest of the PC market? Also not counting porting how much time would be saved in testing, performance optimization, and stability testing if mantle was the only thing the developers cared about?

 

[blastingcap]

Paul only time will tell.

 

[PPB]

People also forget that Mantle enabled games wont mean in a compromise in optimizations towards DX. This also enables the dev to spend 0 time optimizing GCN for DX in a mantle enabled game (because they will be using mantle anyways) and spend all his DX optimizing time targeting only NV and Intel.

 

[frozentundra123456]

The reason people are having difficulty understanding this "simple fact" is that more devs keep signing up for Mantle then wax lyrically about how awesome it is.

It's now been added to what, 4 or 5 game engines? Considering it's not even in beta yet that's unreal. Pun intended.

This myth just keeps coming back again and again. There is no way Dx is going to disappear anytime soon. No matter how easy mantle is to implement, coding for 2 api's will take more time and resources than for Dx alone. As was hinted, the only way it can take less is if resources are diverted away from DX. Even though I have an amd gcn card, I do not view this as a good thing, since it will penalize gamers who do not have a gcn card.

It is simple math, there is no way A(DX) plus B(mantle) is less than A. Simple.

I am not even saying mantle might not be worth it. Nobody really knows for sure until we see real world results. But to say it will save time and money is just wishful thinking at best.

 

[krumme]

People also forget that Mantle enabled games wont mean in a compromise in optimizations towards DX. This also enables the dev to spend 0 time optimizing GCN for DX in a mantle enabled game (because they will be using mantle anyways) and spend all his DX optimizing time targeting only NV and Intel.

Those confused people with bad memory is that the developers running in hordes towards mantle while they curse DX along the way?

 

[SiliconWars]

This myth just keeps coming back again and again. There is no way Dx is going to disappear anytime soon. No matter how easy mantle is to implement, coding for 2 api's will take more time and resources than for Dx alone. As was hinted, the only way it can take less is if resources are diverted away from DX. Even though I have an amd gcn card, I do not view this as a good thing, since it will penalize gamers who do not have a gcn card.

It is simple math, there is no way A(DX) plus B(mantle) is less than A. Simple.

I am not even saying mantle might not be worth it. Nobody really knows for sure until we see real world results. But to say it will save time and money is just wishful thinking at best.

All those recent Gaming Evolved titles that had extra work put in them...no need for that anymore, just go with Mantle instead in future. So Mantle simply substitutes AMD's current GE program. That's a lot of high profile titles going on the past year.

One of Mantle's main strengths is making lower end hardware much better. In fact, that's probably it's biggest draw to devs. They need to look at it themselves and decide if better quality games on lower end hardware is worth the extra effort to them. Certainly in terms of cost it will be worth it, and cost is basically the same as effort for most of the larger devs. If people can play Crysis 4 smoothly on an AMD APU, it's a huge win for them and the 2 man months dev time would be nothing in comparison. It's not about the effort alone, it's about what they get back for it - financially it is assuredly a lot more than what the cost of a Mantle port would be.

 

[FiendishMind]

G-sync is not special; that can be all done actually threw displayport.

NVM, off-topic.

I can think of a few AMD lock-outs such as XGP and CTM.

 

[bystander36]

G-sync also requires a special scaler in the monitor itself.

 

[PPB]

Those confused people with bad memory is that the developers running in hordes towards mantle while they curse DX along the way?

You didnt understand what I said, I said that nobody should fear for less optimized DX versions of a mantle enabled game, because mantle allows the developer to forget about optimizing the DX version for GCN cards (because mantle will take care of those) and focus their dev time into optimizing the DX version exclusively for NVs and Intels GPUs.

 

[bystander36]

You didnt understand what I said, I said that nobody should fear for less optimized DX versions of a mantle enabled game, because mantle allows the developer to forget about optimizing the DX version for GCN cards (because mantle will take care of those) and focus their dev time into optimizing the DX version exclusively for NVs and Intels GPUs.

Do you honestly think that using Mantle automatically optimizes a game?

 

[krumme]

You didnt understand what I said, I said that nobody should fear for less optimized DX versions of a mantle enabled game, because mantle allows the developer to forget about optimizing the DX version for GCN cards (because mantle will take care of those) and focus their dev time into optimizing the DX version exclusively for NVs and Intels GPUs.

What is their incentive to do that then? With gcn and mantle like api all over they cover perhaps 90% of the market for a new aaa performance oriented game?

What is their incentive to try to get the last 30-40 percent performance from dx.

The motivation can only be money from nv. And the one working then will not be the most creative or brightest in the industry. Damn hard work with and old complicated api.

Go see the apu13 q&a and lisnt to the devs arguments. Then you can hear their motivation for using dx.

I think; have a good dev worked with mantle they are not going back. Because mantle gives them control. Mantle puts the good dev in the drivers seat. In a new race car. There is no way they are then going back to the backseat of an 13 year old toyota even if they are higher paid. Its a loss of work identity going back.

 

[PPB]

Do you honestly think that using Mantle automatically optimizes a game?

Non sequitur right there.


Working on a mantle enabled version takes time (as my post implies), that time is not an extra effort since it doesnt make any sense to work hard on a DX optimized version for GCN cards, so current optimizations under DX versions for GCN cards should be scrapped and that time saved is used for the mantle version instead.

Also, this leaves the field for a DX version of the game purely focused on 2 different hardware solutions, and not 3 (NV/Intel instead of NV/Intel/AMD). Mantle is a good thing for everyone, even people who wont really use it.

 

[bystander36]

Non sequitur right there.


Working on a mantle enabled version takes time (as my post implies), that time is not an extra effort since it doesnt make any sense to work hard on a DX optimized version for GCN cards, so current optimizations under DX versions for GCN cards should be scrapped and that time saved is used for the mantle version instead.

Also, this leaves the field for a DX version of the game purely focused on 2 different hardware solutions, and not 3 (NV/Intel instead of NV/Intel/AMD). Mantle is a good thing for everyone, even people who wont really use it.

I don't know if you are just dreaming, naive or what. There will be little to no time saved on not optimizing for GCN cards, as in DX, they would never have used GCN features to begin with. There are many different levels of optimizing, even then, most are not hardware specific.

 

[PPB]

What is their incentive to do that then? With gcn and mantle like api all over they cover perhaps 90% of the market for a new aaa performance oriented game?

What is their incentive to try to get the last 30-40 percent performance from dx.

The motivation can only be money from nv. And the one working then will not be the most creative or brightest in the industry. Damn hard work with and old complicated api.

Go see the apu13 q&a and lisnt to the devs arguments. Then you can hear their mltivation for using dx.

You still dont get it and your evident desire to defend mantle at all costs doesnt even let you agree with someone stating what is actually a favourable point towards implementing mantle.

Ima just use an example, if you still cant get it, the we have a big trouble regarding reading comprehension:

The dev behind X game currently has only a DX version of it. That means that for this dev he has to spend his time optimizing this DX version to work at acceptable levels with AMD/Intel/NV's GPUs.

Now after hearing the benefits of mantle, this dev decides that the next game he will release, game Y, will have a mantle enabled version out of box, alongside with the usual DX version.

His dev time spent on optimizing for all hardware hasnt become longer because of this, even tho he has to put time into a new API enabled version of game Y. Because now all GCN cards have access to mantle, he can scrap the dev time used on optimizing the DX version for GCN and, for the DX version of the game, even NV and Intel would see better optimizations as now the DX version has to play well with only those 2 uarchs. Why would you care if a game runs terribad on a GCN under DX if you would use the mantle version anyways? Dude A running a NV card benefits from this, dude B running a Intel's iGP benefits from this, and obviously, dude C will get the most benefit as he will be running the mantle version with his GCN card.

 

[PPB]

I don't know if you are just dreaming, naive or what. There will be little to no time saved on not optimizing for GCN cards, as in DX, they would never have used GCN features to begin with. There are many different levels of optimizing, even then, most are not hardware specific.

If optimizations werent barely hardware specific, then we wouldnt see disparities running the same game on different hardware, bug wise. Gaming evolved and TWIMTBP programs wouldnt exist in your dream world because of this too.

 

[bystander36]

If optimizations werent barely hardware specific, then we wouldnt see disparities running the same game on different hardware, bug wise. Gaming evolved and TWIMTBP programs wouldnt exist in your dream world because of this too.

I did not say there was no hardware specific optimizations, and there are many levels of optimizations. Only some are hardware specific. The differences between the non GCN AMD cards and AMD GCN cards don't require all new optimizations either. AMD doesn't change everything about their cards from generation to generation.

Don't forget, if a game uses features that favor one brand or another, it will run better on one brand. That doesn't mean they had to optimize it as such. They just had to knowingly use those features.

 

[3DVagabond]

Or maybe the devs will just do less optimizing on the old and broken DX while doing more on Mantle instead.

So now they are going to abandon the overwhelming majority of the market to cater to that little slice of the pie that's GCN? All this before we've even seen a performance demonstration.