The AMD Mantle Thread

[cmdrdredd]

This seems worrisome:



At the end of http://techreport.com/review/25683/delving-deeper-into-amd-mantle-api/3

I said this before. You would have to rewrite large portions of code specifically for mantle. So costs go up and development time rises slightly depending on the base they started with. I dont believe many studios will be able to dedicate that kind of time simply for "extra performance". The benefit to them is zero when they can optimize a game engine for DX11 and it can run good frame rates anyway. I guess if you wanted to think about it a different way, getting extra performance from lower end GCN and APU parts can be a benefit to a developer if it translates to extra game sales. I just wonder if the percentage of people who would be interested or know enough about this and have lower end GCN parts is large enough to translate into increased sales. My first thought is that people with low end GPUs and running integrated dont care about gaming performance. Those who do will buy a beefier dedicated card.

 

[blackened23]

If Mantle gets into the actual big game engines, not just frostbite but unity, cryengine, id tech, and especially Unreal..

Unreal engine 3/4 is out of the question, Tim Sweeney specifically stated that he is completely against using it (this was stated at a Montreal event). And he's the head developer at epic games, so....that leaves game developers (who are using unreal engine) wasting their development time re-writing signification portions of the code if they want to use Mantle. I don't see a lot of devs doing that with UE3 or UE4.

Cry Engine may be a possibility, I guess. I haven't seen the developers make any recent comments about it @ crytek.

 

[krumme]

This seems worrisome:

At the end of http://techreport.com/review/25683/delving-deeper-into-amd-mantle-api/3

Its excactly the oposite of worrisome.

It was expected from most of us that mantle demanded a new engine to be practically feasible because its a radical change. At least i did. And it was a job that was paid by the consoles so to speak. Building mantle into the engines from scratch was then cheap for the benefit.

Now it turns out you can just replace big chunks of an engine to get at least some benefit.

I dont know if its relevant but it just shows imho how motivated the devs is when they start talking about mantle in the perspective of renovating their old car to get some benefit.

Bf4 on the consoles just shows other devs than dice they need to pull their act together and build a new engine.

 

[blackened23]

Now it turns out you can just replace big chunks of an engine to get at least some benefit.

Yeah, I just can't see this happening, this is not a normal thing to do during game development. These guys already have to pay money out the wazoo to simply USE unreal engine - why would they throw a significant portion of the code out? If they paid 200,000$ USD for using the unreal engine 3, that seems completely absurd. Developers aren't going to do this unless they're paid a ton of money by AMD. As it happens, the developers of Thief were paid. Just IMO.

If Mantle + unreal engine is happening on a large scale, AMD should probably sweet talk Tim Sweeney or something. Although he was dead-set against it at the Montreal event.

 

[3DVagabond]

This seems worrisome:

Finally, adding Mantle support to current game engines, as Nixxes did with the version of Unreal Engine 3 used by Thief, can be a challenge. "Native D3D ports will not magically get much higher performance," explained Katsman. "If you emulate the same system on top of Mantle, you will not get much better performance." Fully optimizing an existing engine for Mantle seems to involve breaking and rewriting some chunks of that engine to take advantage of the new development model. But here again, Katsman believes the performance improvements make the effort worthwhile.

At the end of http://techreport.com/review/25683/delving-deeper-into-amd-mantle-api/3

Why is that worrisome? Because it's not going to magically increase performance? Note where the quotes are from Katsman. the part in red, which might be what else you are considering worrisome, were interjected by the author. That's not a quote from Katsman, as it could appear to be, because it's inserted in between statements made by Katsman. Interesting wording as well, "breaking and rewriting". Breaking sounds bad, doesn't it? If he had just said "rewriting some chunks..." would that have sounded as worrisome?

 

[krumme]

Yeah, I just can't see this happening, this is not a normal thing to do during game development. These guys already have to pay money out the wazoo to simply USE unreal engine - why would they throw a significant portion of the code out? If they paid 200,000$ USD for using the unreal engine 3, that seems completely absurd. Developers aren't going to do this unless they're paid a ton of money by AMD. As it happens, the developers of Thief were paid. Just IMO.

If Mantle + unreal engine is happening on a large scale, AMD should probably sweet talk Tim Sweeney or something. Although he was dead-set against it at the Montreal event.

I dont see it happening either. Its a mess. Lets see what thief brings to the table to see if its worth it for gfx. I doubt it. But i guess thief is for truesound?

Imho the sound is gaming wise and for the experience the most important. Unfortunately i will probably only get 50-80 hours bf4 gaming the next year for all gaming time. But i am looking forward to the reaction to the new sound in thief.

 

[Skurge]

Yeah, I just can't see this happening, this is not a normal thing to do during game development. These guys already have to pay money out the wazoo to simply USE unreal engine - why would they throw a significant portion of the code out? If they paid 200,000$ USD for using the unreal engine 3, that seems completely absurd. Developers aren't going to do this unless they're paid a ton of money by AMD. As it happens, the developers of Thief were paid. Just IMO.

If Mantle + unreal engine is happening on a large scale, AMD should probably sweet talk Tim Sweeney or something. Although he was dead-set against it at the Montreal event.

If a small dev like the thief guys can do it, rebellion and Star citizen. I think it's not to bad. If sweeny doesn't use it and Crytek does. Well then Devs will use cryengine and not Unreal. Why would they when it would be easier to use and give better performance.

 

[3DVagabond]

Yeah, I just can't see this happening, this is not a normal thing to do during game development. These guys already have to pay money out the wazoo to simply USE unreal engine - why would they throw a significant portion of the code out? If they paid 200,000$ USD for using the unreal engine 3, that seems completely absurd. Developers aren't going to do this unless they're paid a ton of money by AMD. As it happens, the developers of Thief were paid. Just IMO.

If Mantle + unreal engine is happening on a large scale, AMD should probably sweet talk Tim Sweeney or something. Although he was dead-set against it at the Montreal event.

Pay Sweeney? Was it at the nVidia event in Montreal where he spoke against it? It's not good to play both sides of the fence. If he's already being paid by nVidia taking money from AMD as well isn't really a good look.

If it is good and it is successful and Sweeney doesn't use it, people will just start using other engines that Mantle is in. You don't always have to spread the wealth around to get people to do things if it's a good idea. Time will tell.

 

[krumme]

Unreal engine 3/4 is out of the question, Tim Sweeney specifically stated that he is completely against using it (this was stated at a Montreal event). And he's the head developer at epic games, so....that leaves game developers (who are using unreal engine) wasting their development time re-writing signification portions of the code if they want to use Mantle. I don't see a lot of devs doing that with UE3 or UE4.

Cry Engine may be a possibility, I guess. I haven't seen the developers make any recent comments about it @ crytek.

Mantle is only relevant for new engines.
Unreal 3 is old stuff. Even frostbite 2 is far, far better.
I dont know what could be called Unreal 4 for marketing reasons if ue3 is the comparison. But ue4 needs to be something at least compettitive with fb2 and near fb3 or to much marketshare would be lost on the consoles. The consoles is here and Tim S can say whatever is needed but he needs to sell games competing with ea and he cant do that with a bad old engine relying on old tech. He could. But now that decision is taken from him.

 

[24601]

If a small dev like the thief guys can do it, rebellion and Star citizen. I think it's not to bad. If sweeny doesn't use it and Crytek does. Well then Devs will use cryengine and not Unreal. Why would they when it would be easier to use and give better performance.

Development studios that wanted the latest bells and whistles already are using Cryengine 3/3.5 over Unreal Engine 3/4. Those who actually want their engine to work without massive development overhead already use Unreal Engine.

It would change exactly nothing in terms of what Development Studios do.

 

[cmdrdredd]

Development studios that wanted the latest bells and whistles already are using Cryengine 3/3.5 over Unreal Engine 3/4. Those who actually want their engine to work without massive development overhead already use Unreal Engine.

It would change exactly nothing in terms of what Development Studios do.

I dont believe UE4 is even final yet is it?

 

[Atreidin]

If I remember correctly, back in the day Unreal could run with OpenGL, D3D, Glide, MeTaL, PowerVR, and software renderers. If any company was willing to use Mantle I would have guessed it would be the one that made that game. Things change I guess.

 

[24601]

I dont believe UE4 is even final yet is it?

Lol, Cryengine 3 isn't even final yet. These engines are all in constant development.
No currently popular game engine is final.

Doesn't stop people from making games with them.

If I remember correctly, back in the day Unreal could run with OpenGL, D3D, Glide, MeTaL, PowerVR, and software renderers. If any company was willing to use Mantle I would have guessed it would be the one that made that game. Things change I guess.

When companies realize that customer support (customers in this case being the game developers) is potentially one of the most expensive parts of the whole cost structure of a leading engine developer, that's what happens. Crytek has only gone mainstream since Cryengine 3.

 

[24601]

EDIT: Way to fail forum software.

 

[krumme]

Lol, Cryengine 3 isn't even final yet. These engines are all in constant development.
No currently popular game engine is final.

Doesn't stop people from making games with them.



When companies realize that customer support (customers in this case being the game developers) is potentially one of the most expensive parts of the whole cost structure of a leading engine developer, that's what happens. Crytek has only gone mainstream since Cryengine 3.

It doesnt make sense to talk Mantle without talking consoles. ...That is unless its drawcall limited games like sc lol. But mantle goal is acording to amd own presentation in the start:

1. More ports
2. Quality ports
3. Speed

That a language devs understand.

 

[krumme]

Bella have a perfect example on why Mantle is needed - and it refers to number 2 in prior post:

http://forums.anandtech.com/showthread.php?t=2357058

 

[SiliconWars]

Crytek "evaluating Mantle, looks good so far".

http://www.youtube.com/watch?v=HHdqmzSU-bo&feature=*********&t=16s

 

[psoomah]

What if your laptop or tablet or even smart phone plus a wireless controller plus a good internet connection is all you needed to play anything, even full blown AAA games on your big screen with a Steam like backend to keep track of everything for you? Or play your games on any device anywhere there's a good enough connection with all the effects and eye candy at that the devices native resolution? That is a ginormous potential market.

Mantle, applicable to and making substantially more efficient, every level of the rendering path, is likely the final piece of the puzzle to make that model workable.

Slide 26 of DICE's APU13 presentation ...

Programmability Explicit Multi-GPU 

Explicit control of GPU queues and synchronization, finally!
‒ Implement your own Alternate-Frame-Rendering
‒ Or something more exotic.. 

Use case: Workstation rendering with 4-8 GPUs
‒ Super high-quality rendering & simulation
‒ Load balance graphics & compute job graphs across GPUs
‒ 20-40 TFlops in a single machine! 

Use case: Low-latency rendering
‒ Important for VR and competitive games
‒ Latency optimized GPU job graph scheduling
‒ VR: Simultaneously drive 2 GPUs (1 per eye)

From AMD's CES presentation ...

AMD Strategy: Cloud Lead in the Four Corners of Gaming
- Content
- Cloud
- Console
- Client
.................
Leading in Cloud through Best-In-Class Gaming Technology

-Key Strategic Partnerships (shows CiiNOW but mentions others in talk). CiiNOW is obviously their primary partnership.

From the CiiNOW website:

The CiiNOW Advantage
Innovative Streaming
Deployment Flexibility
CPU/GPU Virtualization
Cross Platform
End-to-end Solution
Fast Deployment
Multiplayer
Active Spectating
Pause & Resume
..................
Better Streaming — Dynamic streaming with up to 1080p support
Better Encoding — Supports both hardware and software encoding to optimize low-end and high-end gaming
Better Quality — Graphics are crisp and high definition
Better Virtualization — Runs on ANY hardware and is optimized on a per CPU/GPU/Game basis
Better Client — Lower CPU usage reduces minimum client device specifications
Better Latency — Lowest end-to-end latency on the market and significantly lower than consoles
..................

And so on.

Mantle latency and efficiency improvements apply almost across the boards.

This is another piece of the puzzle that clicked into place to further illuminate DICE's "Super excited about Mantle!" slide.

Mantle is very possibly the final latency/efficiency element that pushes cloud gaming into the truly feasible.

Also in my mind as I was saying how widely and speedily I expected Mantle to be taken up by the developers and publishers, though this is the first time I've posted about this angle. Developers that don't incorporate Mantle will be at a competitive disadvantage with mobile devices, laptops, desktops, Steam devices and Cloud Gaming. If I could piece this together from the slide decks and various post event Q&A sessions it's hard to imagine there are any developers or publishers that couldn't also. That's why I consider Mantle a done deal at or near 100% adoption by all the major players.

It's all there in the CES, GPU13 and APU13 slide decks waiting to be considered valid at face value, grasped and then interconnected and fitted into AMD's United Gaming Strategy whole.

 

[krumme]

Crytek "evaluating Mantle, looks good so far".

http://www.youtube.com/watch?v=HHdqmzSU-bo&feature=*********&t=16s

Where is the whole part?

But yeaa, its a done deal from dev viewpoint. They want Mantle, but ofcourse there is a lot more tactics to it on the business side.

But i think it opens an important question.

How can you keep the best developers in your team programming for DX that is a black box, if your big EA competitor Dice already jumped Mantle big time, giving the by far best tool?

I mean its creative highly competent, creative specialist with a passion. They dont want to waste their life with something less meaningfull if there is an alternative.

What happened with Mantle was a disruptive thing. At least no worldclass dev. want to use dx anymore for highperformance programming. DX needs to improve radically to become relevant again. Mantle have changed to picture to a new situation, whatever happens.

 

[krumme]

Mantles future
Techreport part 4
http://techreport.com/review/25683/delving-deeper-into-amd-mantle-api/4

 

[Final8ty]

Guys alot is happening with this!!! The performance is nothing short of amazing. Watch this space!!

http://forums.overclockers.co.uk/showthread.php?p=25445660#post25445660

 

[ocre]

What if your laptop or tablet or even smart phone plus a wireless controller plus a good internet connection is all you needed to play anything, even full blown AAA games on your big screen with a Steam like backend to keep track of everything for you? Or play your games on any device anywhere there's a good enough connection with all the effects and eye candy at that the devices native resolution? That is a ginormous potential market.

Mantle, applicable to and making substantially more efficient, every level of the rendering path, is likely the final piece of the puzzle to make that model workable.

Slide 26 of DICE's APU13 presentation ...

Programmability Explicit Multi-GPU 

Explicit control of GPU queues and synchronization, finally!
‒ Implement your own Alternate-Frame-Rendering
‒ Or something more exotic.. 

Use case: Workstation rendering with 4-8 GPUs
‒ Super high-quality rendering & simulation
‒ Load balance graphics & compute job graphs across GPUs
‒ 20-40 TFlops in a single machine! 

Use case: Low-latency rendering
‒ Important for VR and competitive games
‒ Latency optimized GPU job graph scheduling
‒ VR: Simultaneously drive 2 GPUs (1 per eye)

From AMD's CES presentation ...

AMD Strategy: Cloud Lead in the Four Corners of Gaming
- Content
- Cloud
- Console
- Client
.................
Leading in Cloud through Best-In-Class Gaming Technology

-Key Strategic Partnerships (shows CiiNOW but mentions others in talk). CiiNOW is obviously their primary partnership.

From the CiiNOW website:

The CiiNOW Advantage
Innovative Streaming
Deployment Flexibility
CPU/GPU Virtualization
Cross Platform
End-to-end Solution
Fast Deployment
Multiplayer
Active Spectating
Pause & Resume
..................
Better Streaming — Dynamic streaming with up to 1080p support
Better Encoding — Supports both hardware and software encoding to optimize low-end and high-end gaming
Better Quality — Graphics are crisp and high definition
Better Virtualization — Runs on ANY hardware and is optimized on a per CPU/GPU/Game basis
Better Client — Lower CPU usage reduces minimum client device specifications
Better Latency — Lowest end-to-end latency on the market and significantly lower than consoles
..................

And so on.

Mantle latency and efficiency improvements apply almost across the boards.

This is another piece of the puzzle that clicked into place to further illuminate DICE's "Super excited about Mantle!" slide.

Mantle is very possibly the final latency/efficiency element that pushes cloud gaming into the truly feasible.

Also in my mind as I was saying how widely and speedily I expected Mantle to be taken up by the developers and publishers, though this is the first time I've posted about this angle. Developers that don't incorporate Mantle will be at a competitive disadvantage with mobile devices, laptops, desktops, Steam devices and Cloud Gaming. If I could piece this together from the slide decks and various post event Q&A sessions it's hard to imagine there are any developers or publishers that couldn't also. That's why I consider Mantle a done deal at or near 100% adoption by all the major players.

It's all there in the CES, GPU13 and APU13 slide decks waiting to be considered valid at face value, grasped and then interconnected and fitted into AMD's United Gaming Strategy whole.

SPAM!!!!

post #92
https://semiaccurate.com/forums/showthread.php?t=7604&page=10

1000 LOLZ for the part underlined in purple!!!!!!
.
.
.
.
.
.
.

Anyway, i keep hearing that Mantle is so easy to code for but where is the BF4 mantle version? There are some big issues i see with mantle that no one wants to address, Of course there are more efficient ways to render but you have to ask yourself why the market shifted towards Direct X in the first place. Simple, compatibility. Mantle does not solve this major hurdle. Its a huge deal.

You can buy games today on PC that run on a 9800gtx or an HD4870. You can buy a game 5 years old and still enjoy it on a gtx770 or 280x.
The problem is Mantle introduces more problems than it solves. It is not a solution but an attempt to get a leg up on the competition. It is a short term ploy. It segments and isolates. You cannot have a huge variation of HW with mantle. Its GCN only. What is PC gaming if it is not a huge variation of HW and configurations? This is not a solution but an attempt to push everyone to buy a console wannabe PC with AMD APUs and GCN graphic cards. Why would i want that? When I could just buy a PS4?

The biggest hurdle to Mantle obvious. And it is not a hurdle to AMD at all. They want to isolate everyone to AMD HW for the PC as well and really i dont have any drive to do such a thing. When 80% of the PC market use something other than AMD graphics-When 80% of the PC market use something other than AMD CPUs, I just dont see how anyone really expects Mantle to change anything. It will get about as far as AMD can fund it. Thats about it. However many games AMD pays to be ported to Mantle will be ported. It is extra work and it will only benefit GCN cards. It will never take over, it simply cannot. It is not capable. You need something more like DX to have compatibility over a huge variation of HW.

Why are so many pushing Mantle as the best thing since 3d? Its a bunch of hype made to try to sell GCN hardware. We have no real data just a bunch of hot air. Can we not reserve our extremely strong convictions until there is at least something on the market to go on?

 

[GaiaHunter]

Why are so many pushing Mantle as the best thing since 3d? Its a bunch of hype made to try to sell GCN hardware. We have no real data just a bunch of hot air. Can we not reserve our extremely strong convictions until there is at least something on the market to go on?

I don't know what these developers pushing it are thinking...

 

[Haserath]

Anyway, i keep hearing that Mantle is so easy to code for but where is the BF4 mantle version? There are some big issues i see with mantle that no one wants to address, Of course there are more efficient ways to render but you have to ask yourself why the market shifted towards Direct X in the first place. Simple, compatibility. Mantle does not solve this major hurdle. Its a huge deal.

You can buy games today on PC that run on a 9800gtx or an HD4870. You can buy a game 5 years old and still enjoy it on a gtx770 or 280x.
The problem is Mantle introduces more problems than it solves. It is not a solution but an attempt to get a leg up on the competition. It is a short term ploy. It segments and isolates. You cannot have a huge variation of HW with mantle. Its GCN only. What is PC gaming if it is not a huge variation of HW and configurations? This is not a solution but an attempt to push everyone to buy a console wannabe PC with AMD APUs and GCN graphic cards. Why would i want that? When I could just buy a PS4?

The biggest hurdle to Mantle obvious. And it is not a hurdle to AMD at all. They want to isolate everyone to AMD HW for the PC as well and really i dont have any drive to do such a thing. When 80% of the PC market use something other than AMD graphics-When 80% of the PC market use something other than AMD CPUs, I just dont see how anyone really expects Mantle to change anything. It will get about as far as AMD can fund it. Thats about it. However many games AMD pays to be ported to Mantle will be ported. It is extra work and it will only benefit GCN cards. It will never take over, it simply cannot. It is not capable. You need something more like DX to have compatibility over a huge variation of HW.

Why are so many pushing Mantle as the best thing since 3d? Its a bunch of hype made to try to sell GCN hardware. We have no real data just a bunch of hot air. Can we not reserve our extremely strong convictions until there is at least something on the market to go on?

A 4870 can't run Battlefield 4 because it can't run Direct X 11.

Mantle only supports GCN because right now GCN is the only card with all the features and drivers.

Compatability is supposed to be equal with DX. GCN can run DX11 and Mantle because it was made for both.

Even if Mantle is better, technology politics might kill it. If Intel/Nvidia/AMD could agree to use it, that would mean no need of DX anymore.

 

[SiliconWars]

Anyway, i keep hearing that Mantle is so easy to code for but where is the BF4 mantle version?

It's 5 weeks in to it's 8 weeks of development. Easy to code for doesn't mean instant gratification.

There are some big issues i see with mantle that no one wants to address, Of course there are more efficient ways to render but you have to ask yourself why the market shifted towards Direct X in the first place. Simple, compatibility. Mantle does not solve this major hurdle. Its a huge deal.

And yet devs are probably writing for more render targets than they ever did. Mantle is just another of 5 or 6 that are already being coded for.

You can buy games today on PC that run on a 9800gtx or an HD4870. You can buy a game 5 years old and still enjoy it on a gtx770 or 280x.
The problem is Mantle introduces more problems than it solves. It is not a solution but an attempt to get a leg up on the competition. It is a short term ploy. It segments and isolates. You cannot have a huge variation of HW with mantle. Its GCN only. What is PC gaming if it is not a huge variation of HW and configurations? This is not a solution but an attempt to push everyone to buy a console wannabe PC with AMD APUs and GCN graphic cards. Why would i want that? When I could just buy a PS4?

You'd want it because you already have a PC maybe. Maybe you already have a GCN card, or would like to have better than PS4 performance for $300.

The biggest hurdle to Mantle obvious. And it is not a hurdle to AMD at all. They want to isolate everyone to AMD HW for the PC as well and really i dont have any drive to do such a thing. When 80% of the PC market use something other than AMD graphics-When 80% of the PC market use something other than AMD CPUs, I just dont see how anyone really expects Mantle to change anything. It will get about as far as AMD can fund it. Thats about it. However many games AMD pays to be ported to Mantle will be ported. It is extra work and it will only benefit GCN cards. It will never take over, it simply cannot. It is not capable. You need something more like DX to have compatibility over a huge variation of HW.

That's fine, the people using GCN cards will reap the benefits while the rest make do with their DX compatibility. No problem! :thumbsup:

The tiny cost for a Mantle version means AMD can easily pay for it and more. If it's costs $300K, AMD can pay $600K. They are a multi-billion dollar company, this is small change that they are already paying with GE titles anyway.