AMD's new Mantle API

[erocker]

That's just fine.

Stay on topic though.

 

[Lionheart]

Where did I hate on mantle everyone here wants to jump for joy at a 15-20% performance bump and thats fine, but Mantle as an API doesnt make games better it just means more of the same shit runs better lol. And yes this is totally off topic surprised it took this long for someone to say something.

But tell me what is there to truly be excited about with Mantle?

if a game runs at 100 fps already and it can be pushed to 120 oooo it must be infinitely better right ?

or from 25 fps to 30 fps its gonna change the world of gaming.

A new API is interesting but then Mantle isnt really new. It doesnt make games better it just helps developers get a bit more performance now so they can slap on more shit later lol.

Good example DX11 its far faster than DX10 and DX9 its the inclusion of ever more post process filters that has slowed DX11 down. Go figure. Regardless I Hope Mantle proves to be game changing in terms of performance I really do but I just highly doubt it knowing AMD's track record.

AMD = tessellation 10 + years ago just now becoming mainstream
AMD = multi- gpu is prone to issues
AMD = latency issues in games with their branding on it with multi-gpu making it worse
AMD = takes 6 months to a year to fix said issue its not fully fixed yet
This same company now wants to provide a unified graphics API that allows closer access to the GPU awesome lets see how that plays out

This is coming from someone that has run a Ati 9600 / 4870x2 / 5850 xfire / 6950 xfire / 6970 xfire / 7970. Only ran Nvidia with the 7800 GTX / 8800 GTS and GTX 670. My faith in AMD is lacking do to many past mistakes and screw ups. Its rather hard to believe in an API created by a corporation that has a hard enough time getting products out the door with proper support.

You make a good point but I'm still excited for Mantle, mainly due to the next gen consoles being all AMD hardware based (WiiU as well I guess) & SteamOS, it's like Valve & all the other game developers just want to get rid of all the bloatware that's riddled in Direct X & Windows OS's giving hardware level access to the actual hardware itself... (Hope that made sense)

 

[TheoneandonlyMrK]

You do realize you are arguing with a guy with an education in game design right? He easily has more knowledge on the subject then everyone combined in this thread.

I have spoken to him before and do so with respect and I didn't say he was wrong, I agree , , and it is well worth its own thread .
Read title inject relevance.

 

[crazyeyesreaper]

Again dont get me wrong I HOPE MANTLE IS AMAZING

but so far DICE / Crytek / Unreal Engine are not enough.

DICE is great Frostbite is an EA engine so many AAA games might use mantle this is good.

Crytek is going Free to Play in the future so there games = lowest common denominator so Mantle isnt even needed for them and their game engine has a LONG way to go before it gains any serious traction as the technical documentation is lacking.

and well I already beat Unreal Engine to death.

Mantle can be amazing but game engine support doesnt mean shit.

You need amazing games not casual garbage where mantle uplifts and makes a huge impact. That is what will be key.

 

[TheoneandonlyMrK]

Again dont get me wrong I HOPE MANTLE IS AMAZING

but so far DICE / Crytek / Unreal Engine are not enough.

DICE is great Frostbite is an EA engine so many AAA games might use mantle this is good.

Crytek is going Free to Play in the future so there games = lowest common denominator so Mantle isnt even needed for them and their game engine has a LONG way to go before it gains any serious traction as the technical documentation is lacking.

and well I already beat Unreal Engine to death.

Mantle can be amazing but game engine support doesnt mean shit.

You need amazing games not casual garbage where mantle uplifts and makes a huge impact. That is what will be key.

I just think that given everyones recent love of open standards some interesting base level features could be forgot about allowing for new*complexities*In Ai for eg.
I agree with most of your points regarding lazy and unnecessary development but disagree that an api across console pc and arm handset's could be anything but freeing and good for devs.

If they could release there high budget title across more formats directly and easily then everyone wins and imho mantel hsa opengl/cl cuda stream are all steps along this path.

Mantels main unwritten killer app imho is that it clearly is not just about cpu's and gpu's, its being built from the ground up to be hardware aware on a different level, , all the hardware dsps gpus cpus and other co processors, to me it's what's not yet officially been confirmed that interests.
Devs want better tools and in general easier tools allow more amateur work so another minecaft comes sooner.

 

[newconroer]

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

Fixed that for you .... again.

@Crazy

Your DirectX vs Mantle split performance comparisons are irrelevant. It's like saying Linux removes the middle man creating a more robust and buoyant environment, resulting in better performance but then someone comes along and increases power demands which removes any aforementioned gains in performance.

Yet that IS the point of trying to maximize efficiency. You unlock the ability to achieve more for the same (or less) and then spend your spare resources on new features.

 

[The Von Matrices]

Mantle can be amazing but game engine support doesnt mean shit.

You need amazing games not casual garbage where mantle uplifts and makes a huge impact. That is what will be key.

I agree that better game concepts and stories need to come about, but Mantle (or any game engine or hardware) isn't going to change that for better or worse. If anything, licensed game engines allow studios to dedicate more resources on programming that directly improves the end-user experience (rather than programming the fundamental logic on an engine). If indie developers always had to program their own game engines then such a community wouldn't exist. Look at what Adobe Flash has done for casual gaming on the internet as an example.

What you are arguing is that executives and management have taken the savings on licensed game engines and used them as an excuse to cut down even more on employees designing games. My argument is that you are missing the whole picture; any publicly owned company will do that, whether or not they license a game engine. That is not a reason to dislike licensed game engines, that is a reason to dislike that publicly owned companies focused only upon profit have taken over the video game software market. Support indie developers, but don't argue that the game engines are the reason for the lack of creativity.

 

[bencrutz]

Where did I hate on mantle everyone here wants to jump for joy at a 15-20% performance bump and thats fine, but Mantle as an API doesnt make games better it just means more of the same shit runs better lol. And yes this is totally off topic surprised it took this long for someone to say something.

But tell me what is there to truly be excited about with Mantle?

if a game runs at 100 fps already and it can be pushed to 120 oooo it must be infinitely better right ?

or from 25 fps to 30 fps its gonna change the world of gaming.

A new API is interesting but then Mantle isnt really new. It doesnt make games better it just helps developers get a bit more performance now so they can slap on more shit later lol.

Good example DX11 its far faster than DX10 and DX9 its the inclusion of ever more post process filters that has slowed DX11 down. Go figure. Regardless I Hope Mantle proves to be game changing in terms of performance I really do but I just highly doubt it knowing AMD's track record.

AMD = tessellation 10 + years ago just now becoming mainstream
AMD = multi- gpu is prone to issues
AMD = latency issues in games with their branding on it with multi-gpu making it worse
AMD = takes 6 months to a year to fix said issue its not fully fixed yet
This same company now wants to provide a unified graphics API that allows closer access to the GPU awesome lets see how that plays out

This is coming from someone that has run a Ati 9600 / 4870x2 / 5850 xfire / 6950 xfire / 6970 xfire / 7970. Only ran Nvidia with the 7800 GTX / 8800 GTS and GTX 670. My faith in AMD is lacking do to many past mistakes and screw ups. Its rather hard to believe in an API created by a corporation that has a hard enough time getting products out the door with proper support.

i think you looked it from dev point of view for too much.
if mantle can bring games smoother on slower PC, it's a win. not everyone can afford titan.

i do expect consoles games being ported to PC run smooth on my crappy PC.

and regarding tessellation, amd is a tech company.
it provide tessellation but its up to dev to harness it, just like it provide amd64 for microsoft (and apps devs) to use it, but how long would it take for x64 to actually gain a traction in desktop computing world is not (solely) up to amd.

whatd you expect? them stop innovating? would you rather them never built tessellation feature in gpu, never built amd64 cpu?

game coders may think mantle is just a POS that add complexity to their work/job. but imho, dev should fire lazy ass coders.

 

[crazyeyesreaper]

I agree that better game concepts and stories need to come about, but Mantle (or any game engine or hardware) isn't going to change that for better or worse. If anything, licensed game engines allow studios to dedicate more resources on programming that directly improves the end-user experience (rather than programming the fundamental logic on an engine). If indie developers always had to program their own game engines then such a community wouldn't exist. Look at what Adobe Flash has done for casual gaming on the internet as an example.

What you are arguing is that executives and management have taken the savings on licensed game engines and used them as an excuse to cut down even more on employees designing games. My argument is that you are missing the whole picture; any publicly owned company will do that, whether or not they license a game engine. That is not a reason to dislike licensed game engines, that is a reason to dislike that publicly owned companies focused only upon profit have taken over the video game software market. Support indie developers, but don't argue that the game engines are the reason for the lack of creativity.

Really I see alot of indie developers coding their own graphics engines on a regular basis more than a few guys on the Total War Forums have coded engines with Direct X 11 and Open GL functionality themselves.

The only reason UE is as big as it is is due to the documentation and included items which make things rather simple compared to other options.

Where Did I say I disliked that publicly owned company? Last I checked I didn't and game engines + publishers are a a reason for lack of creativity when regurgitated crap brings in 100s of millions in revenue what reason is their to be creative. Mass market appeal has killed that. Game engines focused on turning out a product as soon as humanly possible with as little extra effort as needed has created a vicious cycle that doesnt look to end anytime soon. Their are only a few game engines Ive seen that truly innovative and show promise and yet most likely they wont get used but for a select few titles.

EPIC regularly turns out UE updates but they also tend to get the most out of their engine with simple things like mouse smoothing being possible to disable etc. In todays gaming industry with budgets ballooning past hollywood movies in cost the corner cutting is ridiculous. But I digress I am bored and tired of trolling in this thread. It has been rather fun though. Everyone has their opinion. Mine is that no matter the performance Mantle brings without full support from Nvidia developers wont use it to push things further it just becomes another switch in the engine for AMD hardware. Much like Shader Model 2.0a and 2.0b for Nvidia and AMD respectively.

All one needs to do is look at the next crop of games example would be Watch dogs = 30 fps at 720p

PC minimum
Minimum requirements:
Supported OS: Windows Vista SP2 64bit, Windows 7 SP1 64bit, Windows 8 64bit
Processor: Intel Core 2 Quad Q8400 @ 2.66Ghz or AMD Phenom II X4 940 @ 3.0Ghz - Nearly 6 year old CPU requirement
RAM: 6 GB - higher than the norm but not outrageous most modern systems have 8gb even prebuilt shit 3-4 years ago did
Video Card: 1024 VRAM DirectX 11 with Shader Model 5.0 Not exactly outrageous 5770 would fall into this category and their dirt cheap.
Sound Card: DirectX 9 compatible Sound Card
This product supports 64-bit operating systems ONLY.

Recommended requirements:
Processor: Core i7 3770 @ 3.5Ghz or AMD FX-8350 @ 4.0Ghz - definitely a big jump up on the surface but many games have been recommending these kinds of CPUs for some time
RAM: 8 GB - this is a unique change and interesting
Video Card: 2048 VRAM DirectX 11 with Shader Model 5.0 or higher - 2gb GPUs have been mainstream for awhile.
Sound Card: Surround Sound 5.1 capable sound card
Supported video cards at time of release:
nVidia GeForce GTX460 or better, GT500, GT600, GT700 series - GTX 460 yea not exactly a high end GPU for a recommended spec
AMD Radeon HD5850 or better, HD6000, HD7000, R7 and R9 series
Intel Iris Pro HD 5200. - Oh look an Integrated solution is capable of recommended. yea

So from what I see Mantle isnt required to eek out optimized performance. If these are the specs of next gen games its not that huge an increase. fact is the weakest system in the minimum spec category will run a better looking higher frame rate version then the older systems as is. So I fail to see the not everyone can afford Titan Argument. As the recomended performance level for this game is around a $750 PC if that which is a far cry from Titan.

And no one will fire lazy coders. As Bill Gates said “I choose a lazy person to do a hard job. Because a lazy person will find an easy way to do it.” you would be surprised how true that really is.

 

[RejZoR]

They are using tessellation completely wrong anyway. Instead of using it to improve performance they use it to churn out insane number of polygons where you don't see any difference anyway, making performance worse with hardly any visual improvement instead of keeping the games look around the same poly count wise, but keep performance very high with properly used tessellation.
Sort of better LOD without that obvious model flipping when you move further away.

And i think that has always been the idea behind tessellation, but some smart ass thought it would be better to use it to pump up poly count for no reason and everyone just followed.
Of course with examples of smooth round objects which we never could see in any real game.
That's why every single new game that ofered tessellation had far worse performance with basically no visual improvements and they had to point them out with crappy screenshots for anyone to even notice it. Fail.

For example i can't forget the AvP examples where i still can't forget the railings that were probably made out of 10 polygons instead of 500, yet they used this poly bump on alien model which is black and always lurks in the shadows. What's the point wasting polygons that way instead of for example that railing which player stops by and just notices it being totally low poly.

Or that rally game that was the first with tessellation. It ran like crap and only improvement was some flags and pools of water. And that was it, where the game ran like crap all the time because of it. Design facepalm...

 

[crazyeyesreaper]

Thats another good point Rejzor The Unigine Heaven Benchmark is a good example of overly tessellated but anyone that watches the wireframe will see everything near the camera has high polycounts while further away gets less and less its dynamic and seamlessly integrated and used in a better way. Characters this doesnt work well with for instance its always better to manually create a character models LODs however, for scenery , trees, props, the enviroment and its LOD levels it makes an amazing difference. As it mitigates the model pop in.

Good examples of this is Trees in say Total War, where the LOD meshes dont even line up as you get closer a tree is replaced by an entirely different model that looks nothing like its other LOD levels. etc Things like this are visible in every game I have ever played some people miss it I however do not.

The only time tessellation needs to be applied to characters is with heavily deforming animated areas so elbows / knees / mouth and face. but again this can be better handled with a higher res mesh for those situations however that work is seldom done. As tessellation does not and cannot take into account facial muscle direction for proper deformation. Fun topic for me bores everyone else I am sure.

 

[bencrutz]

So from what I see Mantle isnt required to eek out optimized performance. If these are the specs of next gen games its not that huge an increase. fact is the weakest system in the minimum spec category will run a better looking higher frame rate version then the older systems as is. So I fail to see the not everyone can afford Titan Argument. As the recomended performance level for this game is around a $750 PC if that which is a far cry from Titan.

well, mantle isn't out yet, so....

youre missing the console port to pc part. am hoping with mantle console port would run smooth on pc.

And no one will fire lazy coders. As Bill Gates said “I choose a lazy person to do a hard job. Because a lazy person will find an easy way to do it.” you would be surprised how true that really is.

youre playing with words now. that define a smart coder, not a lazy one, regardless how gates said it.

 

[BiggieShady]

The only time tessellation needs to be applied to characters is with heavily deforming animated areas so elbows / knees / mouth and face. but again this can be better handled with a higher res mesh for those situations however that work is seldom done. As tessellation does not and cannot take into account facial muscle direction for proper deformation.

Max Payne 3, for example, has tesselated characters that show you how smooth are their ears and collars.

The way Rockstar did the sahader, as I understood it, is that not all polys are teesselated - only polygons that are big enough or appropriately angled are dynamically tesselated ... also a nice way to integrate tesselation into existing production pipeline without huge performance hit - for a thing you don't notice without a screenshot though

What would be more visible in games IMO, is leveraging terrain LOD between parallax occlusion mapping and tesselation for crazy good looking ground surface closeups. Then, of course, you get character feet intersecting with the ground (just like in Dragon Age 2 with tesselation enabled)

Fun topic for me bores everyone else I am sure.

The topic is as fun as topics get

... speaking of topic, to stay on topic, for AMD GCN GPUs, Mantle API will be able to contribute to scene image quality by having more objects on screen, rather than having better looking objects on screen ... and these things are mitigated very successfully these days with advanced occlusion culling algorithms. What remains to be seen is how detailed and densely packed game worlds will be by using Mantle with the advanced culling techniques.