The AMD Mantle Thread

[blackened23]

I made nothing up. AMD included NFS rivals on their presentation slides at GPU14. Turns out NFS rivals isn't getting Mantle. What I assume happened was AMD made a bucket list of all frostbite 3/ EA games assuming that all of the developers would use Mantle. Turns out, that isn't happening? At least with NFS Rivals it isn't.

So that begs the question of what other frostbite 3 games which haven't been confirmed to get Mantle, which ones are getting it and which aren't. I guess it's time for you to dig into that and get back to us, because no other developers have confirmed Mantle for their FB3 games that i'm aware of. Then you have games like PvZ zombie warfare which wouldn't benefit anyway, since the requirements are low in the first place.

 

[Erenhardt]

If developers relocate resources from the DX version to the Mantle version, the DX version suffers under almost all circumstances, otherwise, they'd have just let those resources go.

What if it is not true.

Here is how:
DX "is a black box"
Mantle is easy to tweak and optimize your whole engine/game.
What if, taking some of the people trying to find a way to make an engine/game somewhat performing ok in the black box, and moving them to work out optimizations under dev friendly mantle environment. Then taking those optimizations straight to dx. Without wasting time in dx trying to figure out what is wrong.

 

[OCGuy]

This is from Dice, not AMD. Whether right or wrong, this is not bad AMD marketing.

The engine is in fact "Mantle" ready, but each game still has to specifically be programmed to take advantage of it.

 

[Despoiler]

If developers relocate resources from the DX version to the Mantle version, the DX version suffers under almost all circumstances, otherwise, they'd have just let those resources go. They already don't allocate OpenGL resources in 99% of game development because the additional market isn't large enough to support the investment costs. What makes Mantle any different? The entire market encompassed by Mantle is included in the DX market.

If, at some point in the future, AMD has Mantle support on a vast majority of their marketshare, it might become a strategy to develop against Mantle and DX. And even then, that is a dubious claim.

I understand developers wanting it. I get it. I am a developer. I love new technologies. I wish I could use all kinds of new stuff. However, it makes little financial sense from a business aspect.

Does it make little sense? I think the point is the DX is suffering on it's own. The amount of trade offs developers are having to make and the amount of time they need to put into work around DX's inefficiencies have hit a point where looking into a new API makes financial sense.

 

[tonyfreak215]

AMD is paying now, but can they afford to continue to pay every developer to implement Mantle? If AMD can, and does, then I would agree Mantle is free. If they don't, developers are going to have to provide additional resources or reallocate resources to implement Mantle. And, unless you live in some fairy tale world, publishers aren't just going to be giving developers more money.

AMD is not paying for it! AMD did not have to sell the idea to EA nor DICE. They were approached.

Johan Andersson wanted "Mantle". He pitched the idea of a low-level API to multiple companies; Intel and Nvidia included.

http://www.heise.de/newsticker/meld...ber-AMDs-3D-Schnittstelle-Mantle-2045398.html

DICE and AMD are "Co-creators". AMD could have simply been replace by anyone else....

 

[OCGuy]

AMD is not paying for it! AMD did not have to sell the idea to EA nor DICE. They were approached.

Johan Andersson wanted "Mantle". He pitched the idea of a low-level API to multiple companies; Intel and Nvidia included.

http://www.heise.de/newsticker/meld...ber-AMDs-3D-Schnittstelle-Mantle-2045398.html

DICE and AMD are "Co-creators". AMD could have simply been replace by anyone else....

Oh man give this up. There is nothing wrong with how business works.

 

[tonyfreak215]

What if it is not true.

Here is how:
DX "is a black box"
Mantle is easy to tweak and optimize your whole engine/game.
What if, taking some of the people trying to find a way to make an engine/game somewhat performing ok in the black box, and moving them to work out optimizations under dev friendly mantle environment. Then taking those optimizations straight to dx. Without wasting time in dx trying to figure out what is wrong.

The techniques discovered/practiced for the PS4, will apply towards Mantle.

 

[Paul98]

What I will be looking for is reviews with an i5 and 7950ish GPU.

 

[ASM-coder]

I made nothing up. AMD included NFS rivals on their presentation slides at GPU14.

Can you show that slide?

Then you have games like PvZ zombie warfare which wouldn't benefit anyway, since the requirements are low in the first place.

It says right on the slide that they are targeting the APU market with Mantle support. So they believe Mantle will improve performance on a lesser powered processor such as Kaveri, and they made the business decision that there might be a few bucks to be made by tapping that market.

 

[Spjut]

What I will be looking for is reviews with an i5 and 7950ish GPU.

I hope reviewers will give it a go on Core 2 Quads and the Phenom CPUs with an HD 7950, such tests would really show just how big a difference Mantle can make...I have a personal interest in such a test

I'm afraid most reviewers will just use FX 6300 and FX 8320 and call it a day

 

[OCGuy]

What I will be looking for is reviews with an i5 and 7950ish GPU.

I don't think a Sandy or beyond i5 bottlenecks a 7950 at 1080p or above...

I hope reviewers will give it a go on Core 2 Quads and the Phenom CPUs with an HD 7950, such tests would really show just how big a difference Mantle can make...I have a personal interest in such a test

I'm afraid most reviewers will just use FX 6300 and FX 8320 and call it a day

The amount of time it takes to review even modern hardware makes it tough to spend time on legacy systems.

However, you can approximate what gains you would achieve. A 280x is a 7970 in a new package. 7950 is x% of a 7970, etc etc.

 

[GaiaHunter]

If developers relocate resources from the DX version to the Mantle version, the DX version suffers under almost all circumstances, otherwise, they'd have just let those resources go. They already don't allocate OpenGL resources in 99% of game development because the additional market isn't large enough to support the investment costs. What makes Mantle any different? The entire market encompassed by Mantle is included in the DX market.

If, at some point in the future, AMD has Mantle support on a vast majority of their marketshare, it might become a strategy to develop against Mantle and DX. And even then, that is a dubious claim.

I understand developers wanting it. I get it. I am a developer. I love new technologies. I wish I could use all kinds of new stuff. However, it makes little financial sense from a business aspect.

If Mantle is not financially sound, there will be no resources relocated.
If Mantle is financial sound, then it makes sense for them to relocate those resources.

Again it is a non problem for the developers.

 

[Spjut]

The amount of time it takes to review even modern hardware makes it tough to spend time on legacy systems.

However, you can approximate what gains you would achieve. A 280x is a 7970 in a new package. 7950 is x% of a 7970, etc etc.

Sure, I get that.
I just mean, since Mantle seems to be mostly about much better CPU performance, it's kind of a waste to spend time benchmarking the various CPUs in the higher-end spectrum, when the big gains should be shown by using the slowest CPUs. If for example a Core 2 Quad gains 40% average using Mantle, that would say alot for users of for example Sandy Bridge i5s, alot more than just benching an i5 with a R9 290x IMO

 

[itsmydamnation]

I don't think a Sandy or beyond i5 bottlenecks a 7950 at 1080p or above...

it will be interesting to see, you are always bottlenecks by something and that bottleneck can change 100 times a frame. So what will be interesting to see is if games that dont have good input to photon (to borrow from oculus) latency/jitter but aren't largely CPU bound see improvements.

the other thing is if mantle becomes more prevalent/leave Beta etc. you might see settings that are only allowed/practical to be set when mantle is enabled ( much like physx on BL2 for example).

Sure, I get that.
I just mean, since Mantle seems to be mostly about much better CPU performance, it's kind of a waste to spend time benchmarking the various CPUs in the higher-end spectrum, when the big gains should be shown by using the slowest CPUs. If for example a Core 2 Quad gains 40% average using Mantle, that would say alot for users of for example Sandy Bridge i5s, alot more than just benching an i5 with a R9 290x IMO

it isn't just about CPU, thats just the stuff that low hanging fruit. GPU's already get very good utilization, mantel allows for improvement in efficiency and exposure of things that aren't in DX but that's likely only ever going to be incremental improvements.

example NV have in OGL bindless texutring and people are using that as a you dont need mantle look at this instead, well all texturing on GCN is done bindlessly in hardware ( could very well be for NV to but they dont say) now if a dev want to do everything bindlessly its quite simple with mantel. But with OGL extensions you have different GPU's ( even from the same vendor) supporting different extensions. if your using multipule it can become problematic quickly.

 

[Paul98]

Many of us don't play on ultra or high, but with most of the visual stuff off or on low. Doing this so you can see the enemy better and so the min fps doesn't drop that much. This should mean a good bump in FPS for this usage.

Such as I am running mostly on low, with Mesh Ultra and I think two other settings on medium. I would love to have it where at no point I drop below 60fps, I am almost there as it is. I hope mantle get's me the rest of the way.

 

[ASM-coder]

If Mantle is not financially sound, there will be no resources relocated.
If Mantle is financial sound, then it makes sense for them to relocate those resources.

Again it is a non problem for the developers.

And keep in mind that EA has made a decision to consolidate by using the Frostbite 3 engine company wide.
There will probably be developers and QA staff available that may otherwise be made redundant.

And it doesn't even have to be like this. There will be business plans and market analysis made for every product developed. Developers and QA will be added if there is a buck to be made.

When you ARE the revenue generator, rather than an expense(like most programmers), budgets and manpower are not so fixed.

 

[krumme]

how do you know they are paying

Dice/ea had a challenge with fb2 and now fb3 engine vs eg. ue3 engine that it demanded far more power. Mantle brings fb3 engine to cheap machines thereby removing the most downside side of fb engine.

Ea want to use fb as future platform for all their games. Mantle is simply a prerequisite for that. They would ofcource like to have nv using the same api, but it remains to be seen if either nv or amd wants that.

 

[krumme]

Oh man give this up. There is nothing wrong with how business works.

Oh man try and look at amd balance sheet. The idea that amd is in charge and decides and make the decisions is pathetic. Some dudes at dice and amd get some good ideas but the decision is solely where the money is; at ea.

Fb is one of 3 interesting huge engines and the most important one here. How on earth can anyone think 8m from amd to dice is going to pay for that platform to implement mantle. Argg. Its like women and economy - always at least one 0 is missing. Ea dont give a damn about amd they only do this to beat the other in this entertaintment market.

 

[cmdrdredd]

The techniques discovered/practiced for the PS4, will apply towards Mantle.

And it will only apply to a small percentage of the market.

Oh man try and look at amd balance sheet. The idea that amd is in charge and decides and make the decisions is pathetic. Some dudes at dice and amd get some good ideas but the decision is solely where the money is; at ea.

Fb is one of 3 interesting huge engines and the most important one here. How on earth can anyone think 8m from amd to dice is going to pay for that platform to implement mantle. Argg. Its like women and economy - always at least one 0 is missing. Ea dont give a damn about amd they only do this to beat the other in this entertaintment market.

If EA don't give a damn why then did they not say "stop working on mantle right now. The entire game is a cluster...and it needs to be fixed yesterday"? It's simple...somewhere along the lines money changed hands somewhere.

 

[DamnedLife]

I made nothing up. AMD included NFS rivals on their presentation slides at GPU14. Turns out NFS rivals isn't getting Mantle. What I assume happened was AMD made a bucket list of all frostbite 3/ EA games assuming that all of the developers would use Mantle. Turns out, that isn't happening? At least with NFS Rivals it isn't.

So that begs the question of what other frostbite 3 games which haven't been confirmed to get Mantle, which ones are getting it and which aren't. I guess it's time for you to dig into that and get back to us, because no other developers have confirmed Mantle for their FB3 games that i'm aware of. Then you have games like PvZ zombie warfare which wouldn't benefit anyway, since the requirements are low in the first place.

Dude it said "Ghost games" Developers of NFS Rivals, that slide had NFS logo on it and Ghost games, NOWHERE IT SAID NFS RIVALS specificly.

Future NFS titles will have it. STOP FUD

 

[krumme]

If Mantle is not financially sound, there will be no resources relocated.
If Mantle is financial sound, then it makes sense for them to relocate those resources.

Again it is a non problem for the developers.

Its one of the wonders of competition. It nearly always works and will absolutely work in this situation. If mantle sucks it will fade away if not it will not. No need to put a lot of feeling into that. Lol.

 

[tonyfreak215]

And it will only apply to a small percentage of the market.

Funny thing about markets, they change. Apple is a great example of that....

If EA don't give a damn why then did they not say "stop working on mantle right now. The entire game is a cluster...and it needs to be fixed yesterday"? It's simple...somewhere along the lines money changed hands somewhere.

Perhaps it has to do with Mantle also being DICE's (Johnan) baby? Could also have to do with personnel? You think the same people who incorporate Mantle would also squash bugs?

 

[cmdrdredd]

Dude it said "Ghost games" Developers of NFS Rivals, that slide had NFS logo on it and Ghost games, NOWHERE IT SAID NFS RIVALS specificly.

Future NFS titles will have it. STOP FUD

I'd like you to give us the quote that says "future NFS games will have Mantle support from day one going forward" or similar. If anything, you're making up something that wasn't said anywhere. Yeah the logo was shown and the only NFS game currently using the FB3 engine isn't getting Mantle. That much we know. What you said nobody knows and nobody has said a word.

So funny to read some of the negative comments in here. You can easily expect anywhere between a 5-50% improvement.

You guys are the same people that will overclock, and configure settings for hours to get 2-3 more FPS... and yet AMD is giving you more.. much more.. to you completely free (from the consumers perspective) and you complain.

I love the internet.

To get maximum benefit it also is requiring a card that is $150+ overpriced thanks to mining tax. So from my perspective I don't think it matters what AMD is giving anyone.

Funny thing about markets, they change. Apple is a great example of that....

AMD is not Apple...nice try. Who is really gonna pay $700 for a 290x that only has an advantage in one game? Only miners seem to be willing to pay the costs.

 

[smackababy]

Funny thing about markets, they change. Apple is a great example of that....

Are they? Interesting you use Apple as an example to help show why having increased marketshare in the PC market equates to more games being available on your platform.

 

[Phanuel]

To get maximum benefit it also is requiring a card that is $150+ overpriced thanks to mining tax. So from my perspective I don't think it matters what AMD is giving anyone.

And so for that you'd burn AMD at the stake for a free market working how a free market works in response to a pseudo currency demand completely outside of their control?