The AMD Mantle Thread

[Lonbjerg]

Then stop replying because you're talking about something else. I'm not talking about PR.

If you cannot distinguish this situation and the natrue of it, it's your problem.

Your opinion does not matter, neither does mine, reality won't change because you realize that or not.

Whatever you say, I provide the most accurate analysis I can, If you found a better one, then show it.

I don't discuss PR, you can make a thread and form an analysis of Mantle on the basis of it all being fake and a hoax, which I wouldn't be a part of, and you can provide analysis on those points of all the details you found that would make up a case for it being all a joke, I could have done that as well but I haven't choosed to do so, the data in my analysis was insufficient to have any sense to pursue that path.

All Mantle is right now IS PR.
No independant reviews.
No downloadable demoes.
Just canned tech demos from devs pushing their own game-demos.

Feel free to prove me wrong.

 

[mikk]

Bs. There is no way it can be 50% faster on a 6 core cpu and you know it.

Can you explain it or do you like bullshit posts like this? If you are unable to explain better shut up. You're delusional to think they are doing all the work for such a tiny user base because of a 20% speedup and it seems 2 months of work was a bit off. Not only they have to work 2 months for implementation (which is questionable now), they also have ongoing maintenance costs and work when they support another API. Andersson told Mantle in Battlefield utilizes all 8 cores. Under DX11 the gain from 4 cores to 6 cores is pretty bad in Battlefield 4. So from the core scaling alone they surely don't expect just 20%. Not to mention the lower overhead and higher drawcall limit.

You keep bring dx11mt to the table. This is a perfect example of why dx have bad cost/benefit. It doesnt matter what is possible with dx if it takes tons of ressources to get minor benefit. That goes for all the performance aspects of dx. Its simply not suited for performance games in a market with a handfull of engines driving all performance games. Dx doesnt fit the market for performance games - and that more or less includes the consoles.

DX has a bad cost/benefit, really? How bad is Mantle then? What the hell is a performance game? What is your source for your tons of ressources to get a minor benefit? Nvidia, AMD, Intel have debug and performance analyzer tools where render coder get minor performance gains (and bigger gains as well) out of it from hours to days of work. It sounds like there is no optimization work possible for DX11.

You think DICE doesn't know how to implement DX11 properly already ?

I referred to DX11 with command lists. Dice doesn't support multithreaded DX11 rendering over command lists. Andersson himself presented it in a Frostbite speech 2 years ago but as far as I know they don't use it in Battlefield 3/4.

 

[SiliconWars]

Can you explain it or do you like bullshit posts like this? If you are unable to explain better shut up. You're delusional to think they are doing all the work for such a tiny user base because of a 20% speedup and it seems 2 months of work was a bit off. Not only they have to work 2 months for implementation (which is questionable now), they also have ongoing maintenance costs and work when they support another API. Andersson told Mantle in Battlefield utilizes all 8 cores. Under DX11 the gain from 4 cores to 6 cores is pretty bad in Battlefield 4. So from the core scaling alone they surely don't expect just 20%. Not to mention the lower overhead and higher drawcall limit.

It depends on how much work there is for the CPU to actually do. If there is only enough work for 4 cores then having 8 cores is only going to get you a benefit from turbo etc.

As we know from a previous article, DICE has said that Mantle basically completely equalises the CPU's - http://www.techradar.com/news/compu...after-nvidia-graphics-with-kaveri-apu-1199636

"What I have heard from DICE is that what [Mantle] does with the discrete card is it equalizes the CPUs," Kozak explained. "It was only a couple of frames faster before because the CPU doesn't really play into things like that, but [Mantle] eliminates any gap. And essentially it does that by allowing the graphics card to do more, so it becomes the bottle neck."

I'm pretty sure you know this and are now just spreading FUD by attempting to project far higher performance from the CPU gains than are likely.

The alternative to higher fps that can't be got from the engine is that more work is created to utilise more cores, ie with drawcalls. This will make the game look better but not necessarily perform faster.

In short, your 50% faster projection from an increase in CPU cores is total BS and you well know it.

 

[Lonbjerg]

In short, your 50% faster projection from an increase in CPU cores is total BS and you well know it.

How does that differ from your claim about R290X = SLI Titans in BF4 when using Mantle?

 

[SiliconWars]

How does that differ from your claim about R290X = SLI Titans in BF4 when using Mantle?

Because - incredible though this might seem - games actually make use of GPU's more than CPU's.

I know, really hard to believe right?

 

[144HzGamer]

Why is it that this particular thread has attracted quite a few new accounts, all arguing in favor of Mantle's marketing slide greatness?

Just something I've noticed.

And I pose a question: If Mantle is so great, why now at the final hour of when it's BF4 target release have we heard NOTHING about it. 2 man months? Hmm... Perhaps not.

So obviously that was directed at me, oh I don't know... Maybe because it's one, if not the most active Mantle thread out there right now and I wanted to chime in? But I guess the most logical guess would be that I'm being paid by AMD to promote Mantle.

Oh, I suppose I didn't hint enough at the fanatical nature of their posts on a fresh account. I just find it strange that someone makes an account at a forum to post fanatical things about what is currently vaporware.



People keep saying BF4 "has been a mess", but I've not experienced it as such. The launch was a bit worse, but after the first week, it has been smooth sailing. I think quite a lot of people complaining about BF4 are the ones who don't have it, but that is just conjecture on my part.

You would think there would be some Mantle news from any of the Frostbite team. We've heard nothing except "these games are going to use Frostbite (at some point... new Mass Effect?) and that Frostbite will have Mantle in it. We haven't heard anything else, from DICE.

Fanatical posts? have you actually read my posts? if that's the case then you have some serious learning to do.

In fact I actually agree with your point regarding DICE being completely silent about the Mantle update, if it's delayed then make it public, if you guys (DICE & Frostbite team) have learned something about the release of Battlefield 4 is that rushing a product is just wrong on so many levels, we would have preferred a delay over downloading a broken-ish game at launch.

That's one of the reasons why I'm (still) holding out hope for a Mantle release before new years, the silence is just too weird, Johan is a pretty active guy on both twitter and on some forums (Beyond3d, etc), so my gut tells me that if its in fact delayed he would have address it already. That's my opinion.

But it does have significant downsides, in terms of compatibility. The difference between Mantle and DirectX/OpenGL is actually much deeper than its just another API. Mantle's entire design goal is to be fast to run by representing the hardware, pushing hardware limitations into the software. DirectX is actually about 3D models and worlds and how they are put together, you can run it in software on the CPU and it will work just fine. OpenGL equally is about putting together 3D worlds. These abstractions based on the essence of 3D graphics last a lot longer as a concept, because these abstractions aren't about just layering away from hardware they are about making 3D worlds and rendering them with rasterization.

Mantle would not only introduce a completely different concept into this world but do so tied to a single manufacturer and model of card. It throws away decades of careful abstraction to allow hardware/software changes for the future in favour of performance in the short term, only to become a throw away API that when architecture and hardware changes causes changes in the future. I don't doubt the performance benefits might be significant, but its not a good trend for yet another API of any type to be introduced, its certainly not good for an API of this type to be introduced, its the sort of API only aimed at experts.

Just as an aside the problem with Arma 3 is that its single threaded. DirectX calls do take up a significant part of the games thread but it also manages to offload to extra threads (6 in all) during the render process. Its mostly dominated by its simulation updates, its that which in multiplayer seems to grow to a high amount and make the game CPU limited. The rendering process pretty much always achieves about 85 fps on its own and at all times, its just the simulation that seems to get out of hand and I am hoping with my profiles BIS can work out why. Its just an old engine with new spit and polish on the top but its not designed for many core CPUs as a game at all, which considering the complexity of the game is a real problem.

BF4 - the issue I have had with the game are bugs in the gameplay. But graphically it runs really well 110-170 fps for me in 64 player games. Its actually one of the best running games I have ever owned.

I'm not sure I follow...

How is Mantle different from Direct3D/OpenGL in regards to developing "3D worlds, models, etc"?, I'm not trying to come off as disrespectful but you sure look like you have no idea what your talking about mate.

And btw, Mantle is definitely not something that's being developed with a short term vision in mind, in fact, Mantle is being built with a close eye on adding support for future AMD architectures and for other architectures from different vendors (there's even talk about Mantle going mobile), so this is not something that will work on GCN cards and then it'll disappear.

And one last thing, all of the stuff that has been said about Mantle here can be said about NVIDIA's G-Sync, and how it's something that just works on Kepler cards and that it needs a TN panel, etc etc, that's just the start for that particular tech, and that's perfectly fine IMO, I have an AMD card and I love the idea behind G-Sync, I LOVE IT, the fact that it'll push the hardware industry forward excites me, same thing with Mantle, if it lives up to expectations, only good things can come off of it, so the fight against Mantle is definitely something I don't understand.

Now if you absolutely do not believe the developers or the "vaporware PR slides" or whatever you wanna call it, that's a whole different issue. In the case of Johan Andersson, it wasn't all that long ago that he was seen at NVIDIA conferences regularly, hell he was even promoting G-Sync pretty hardcore during NVIDIA's last conference, he just likes innovation wherever he sees it.

 

[Stewox]

I was wondering about that. On the page he says Maxwell isn't coming until Q4 2014, that sentence had a link and when I clicked the link it went to a page where the article was behind a paywall, which I though was strange and not sure what to think about it. I've never seen that before on a tech site. So you think this Charlie Demerjian is valid and has good information?

It's only been a paywall for one year, and he did that to preserve the quality, he would have to close down the site or sell it, or be acquired by another company that would eventually lower the quality of the site in order to make more drama-hit articles which would be written in a fashion to attract more hits and therefore more advertisment revenue. He refused to sacrifice quality and the soul of the site.

All Mantle is right now IS PR.
No independant reviews.
No downloadable demoes.
Just canned tech demos from devs pushing their own game-demos.

Feel free to prove me wrong.

• 1. That is your opinion
• 2. Irrelevant (I'm analyzing current data)
• 3. Irrelevant (I'm analyzing current data)
• 4. My analysis and data shows it's not a canned demo.

To present this in a clearler way:


So what basically I focus on:

• Analyzing Current Data

What I do not focus on:

1. C. PR conspiracy theory
2. C. Waiting for downloadable demo
3. C. Waiting for independent reviews
4. C. What Nvidia's going to do

I do not discuss those 4 just above topics, because I choosed not to do so. I am not interested in those topics objectively and personally. If my early data and analysis found that Mantle was a PR hoax, then I would have stopped right there without discussion any further and would erase it from my head, wouldn't be having this conversation.

In order for me to be objectively interested in those 4 C-points, I would had to be paid for it. The 4th C-point would be of good interest to investors, but I am not a paid market analyst.


My Personal Opinion:
Those first 3 C-points are the easiest which require little work, basically no brain power and very little thinking.

I'm not interested to use my time for posts like "let's wait for reviews", I am not a bored avreage-gamer, my time here is all scheduled.

I don't need reviews, I can make my own mind up, faster and better.

 

[mikk]

It depends on how much work there is for the CPU to actually do. If there is only enough work for 4 cores then having 8 cores is only going to get you a benefit from turbo etc.

Regarding BF4 there must be more than enough work existing, otherwise Andersson wouldn't say perfect parallel rendering - Mantle utilizes all 8 cores.

As we know from a previous article, DICE has said that Mantle basically completely equalises the CPU's

Which is unrealistic with only 20%.

In short, your 50% faster projection from an increase in CPU cores is total BS and you well know it.

Considering that I see up to +20% in an unoptimized DX11MT over DX11 in Pcars I expect something better from a low level API with much more advantages. Unlike DX11MT Mantle is low level>no driver overhead, reduced latency, much easier CPU scaling, drawcall limits much higher and reduced drawcall overhead. All the talk about the bad Directx performance wouldn't make sense if Mantle can do just 20% better.

 

[DiogoDX]

If people bother to watch Jonah presentation he says that driver multithread can do only 2-3 threads and they are aiming to deliver 4 threads with first mantle interation.

 

[Stewox]

Also this, relevant to my big post from earlier:

"It'll blow your socks off in terms of CPU Utilization"
http://www.youtube.com/watch?v=r7Q_nAnnrnI&t=5m50s
(listen for 2 mins or 3)

One thing is GPU and the other thing is CPU

It would be good if you guys explained more clearly which one you mean and It's a bad idea to use combined "overall" predictions anyway.

If people bother to watch Jonah presentation he says that driver multithread can do only 2-3 threads and they are aiming to deliver 4 threads with first mantle interation.

Thank you

I missed that, I'm sure I've heard it "4 core support initially", but I missed the part he actually mentions multi-threaded drivers.

I will use headphones and listen again to catch it. Audio's really a pain in the leg in situations like these.


See, my data is not perfect, but they, reality's even better than I said.()

 

[Lonbjerg]

• 1. That is your opinion
• 2. Irrelevant (I'm analyzing current data)
• 3. Irrelevant (I'm analyzing current data)
• 4. My analysis and data shows it's not a canned demo.

To present this in a clearler way:


So what basically I focus on:

• Analyzing Current Data

What I do not focus on:

1. C. PR conspiracy theory
2. C. Waiting for downloadable dem
3. C. Waiting for independent reviews
4. C. What Nvidia's going to do

I do not discuss those 4 just above topics, because I choosed not to do so. I am not interested in those topics objectively and personally. If my early data and analysis found that Mantle was a PR hoax, then I would have stopped right there without discussion any further and would erase it from my head, wouldn't be having this conversation.

In order for me to be objectively interested in those 4 C-points, I would had to be paid for it. The 4th C-point would be of good interest to investors, but I am not a paid market analyst.


My Personal Opinion:
Those first 3 C-points are the easiest which require little work, basically no brain power and very little thinking.

Just because you type something, dosn't make it true.
Like I said....lots of empty PR from you.

Wake me up when you got more than empty words with no documentation.

I mean, it's not like anyone on IT would ever lie...right?
http://www.youtube.com/watch?v=gLgb9AdnaBI

They promised GPU physics in 9 years ago.
Empty PR dosn't rock my boat.

 

[VulgarDisplay]

Honestly if you aren't here to discuss mantle in a positive manner than stop thread crapping. Go start a "mantle sucks because amd..." Thread and keep your pollution elsewhere. This thread used to be about the positive changes mantle claims to bring to PC games and it is now devolved into a select few professional trolls baiting people incessantly. It doesn't help that people keep responding to them. They aren't going to see the light... Trust me.

Mission accomplished this thread has been flushed down the toilet. Nothing to read in the last 3 pages that isn't complete nonsense.

 

[SiliconWars]

Regarding BF4 there must be more than enough work existing, otherwise Andersson wouldn't say perfect parallel rendering - Mantle utilizes all 8 cores.

Utilizes doesn't have to mean fully utilizes though - 4 cores can do the work of 8 cores, they just need twice the MHz.

Considering that I see up to +20% in an unoptimized DX11MT over DX11 in Pcars I expect something better from a low level API with much more advantages. Unlike DX11MT Mantle is low level>no driver overhead, reduced latency, much easier CPU scaling, drawcall limits much higher and reduced drawcall overhead. All the talk about the bad Directx performance wouldn't make sense if Mantle can do just 20% better.

If there was enough work to fill 8 fully loaded CPU cores of work then yes we would see some incredible gains over quad cores. I'm pretty sure that work just doesn't exist in the DX BF4 implementation so it will have to be created to make use of it in Mantle.

Either way AMD basically can't lose out of this, for the following reason -

If there is enough work to fully load 8 cores then PD is going to look great - even the quads will look great vs the i5.

If there isn't enough work then the CPU's will perform the same, 4 cores to 8. This is still good for AMD but gets even better when you consider Kaveri + asymmetric crossfire *will* perform as top CPU with ease. I believe this option is even better than the game scaling to 8 cores, as Kaveri will launch to being the best gaming CPU for BF4.

Think about this -

Kaveri's 512 GCN shaders are capable of providing around 10%-15% extra performance of the 290X alone.

This could also be the reason for the "delay" in the Mantle patch - they can't release Kaveri benchmarks until the Kaveri NDA is ready to lift anyway.

We know the game doesn't scale great beyond 4 cores.

We know repi has been working on Kaveri recently.

There are a couple of other indicators I've noticed as well, but for me it's pretty clear that Kaveri + 290X is going to beat any Intel CPU + 290X, and will completely annihilate any system with an Nvidia GPU.

 

[blackened23]

Honestly if you aren't here to discuss mantle in a positive manner than stop thread crapping. Go start a "mantle sucks because amd..." Thread and keep your pollution elsewhere. This thread used to be about the positive changes mantle claims to bring to PC games and it is now devolved into a select few professional trolls baiting people incessantly. It doesn't help that people keep responding to them.

Since when are opinions not allowable in a given thread, give me a break please. Trolling, yeah, sure, there are trolls on both sides in this thread. That's pretty obvious - I just read pages of people predicting the death of nvidia. Pretty sure that's trolling because it's Based on TWO MANTLE GAMES CONFIRMED for 2014. And while nvidia has a commanding dGPU market share lead. But i'm with you on this. Trolls should go away. Yet they're on both sides.

The suggestion that only "positive" opinions of Mantle are allowed in this thread is downright laughable. We're here to exchange opinions and thoughts on Mantle. Not everyone here takes AMD marketing or whoever the next JF-AMD is, at face value. If anything, we've learned to never take AMD marketing at face value.

I've said it several times, I hope Mantle does well. Yet reading some of the pro-Mantle stuff in this thread is so over the top it's quite amusing. Whatever though. I'm as anxious as anyone else to see what Mantle does in an actual game that we can play. Pretty sure most folks are done caring about what any marketing slide says, especially those from AMD and their shady nonsense marketing, so hopefully BF4 with Mantle is released soon. That's the only thing that will truly end the nonsense. Whenever AMD decides to put or shut up and show us what Mantle has the potential for in BF4. So like I said....until it's released...who cares about marketing fluff/nonsense.

 

[Stewox]

I mean, it's not like anyone on IT would ever lie...right?

I can say that for Carmack, he's the one who kept talking about how PC is inefficient how it would be nice to have low level API for PC, all that talk, appeared like he was doing something to make that happen turns out he wasn't at all, and now on top of that he's clearly more interested in Oculus Rift which is great but I don't focus that much on it anymore until I can actually afford it hah (and would need a much better GPU also)

Now he's talking against Mantle like it's not something everyone was waiting for, his opinion, but I won't listen to him anymore about it, he's not developing engines anymore.

He didn't exactly lie, so maybe it was a bad example.

http://www.youtube.com/watch?v=gLgb9AdnaBI

Which is a totally different situation and circumstances. I think it would take me quite some time to go all over the points. Like 20 points or something out of my head.

 

[desprado]

Honestly if you aren't here to discuss mantle in a positive manner than stop thread crapping. Go start a "mantle sucks because amd..." Thread and keep your pollution elsewhere. This thread used to be about the positive changes mantle claims to bring to PC games and it is now devolved into a select few professional trolls baiting people incessantly. It doesn't help that people keep responding to them. They aren't going to see the light... Trust me.

Mission accomplished this thread has been flushed down the toilet. Nothing to read in the last 3 pages that isn't complete nonsense.

Stop posting BS all the time.We all are posting facts and not predicting like others.This topic dont say just post PR and positive talks and if u cannot tolerate than leave this topic go to AMD forum there u can see many positive talks

 

[joyzbuzz]

We know repi has been working on Kaveri recently.

"When you are trying to optimize in the sub millisecond range, PCI-E is really a limitation. Async GPU DMA to the rescue!"

I'm not sure I understand this. Does this mean he has figured out a way to get around the PCI-E latency problem using Mantle with Kaveri and a discrete GPU?

 

[joyzbuzz]

Mission accomplished this thread has been flushed down the toilet. Nothing to read in the last 3 pages that isn't complete nonsense.

I am learning things even if you aren't.

 

[joyzbuzz]

I sure don't envy the moderator of this thread.

 

[VulgarDisplay]

All the facts in this thread from both sides are all purely opinions from people who basically have zero knowledge of game design, and how games utilize CPUs and GPUs.

There are a couple people on this forum who I believe have some good knowledge of the topics at hand. They are not participating in this thread however. If they are then I guess I have been hoodwinked because of all the ill formed and nonsensical posts bordering on complete irrelevance I have read in the last three pages.

I also don't care if people have negative things to say about mantle in the thread. I do take issue with the fact that they make no sense and have no relevance to the mantle thread.

You'll all notice how I'm quiet in the nvidia game works thread. It is because I try desperately to live by a simple rule which is if you do not have anything good to say do not say anything at all. I try and stick to that but the trolls in this thread are ruining one of the few interesting threads active in the forum...

 

[Lonbjerg]

Like I said...right now Mantle is PR and vaporware...no matter how "opinionated" people try to talk it up.

And this is not the "All praise - Mantle - Critique is NOT welcome" thread FYI.

Don't like have PR called PR...well...then present something else than PR.

 

[Stewox]

Like I said...right now Mantle is PR and vaporware...no matter how "opinionated" people try to talk it up.

And this is not the "All praise - Mantle - Critique is NOT welcome" thread FYI.

Don't like have PR called PR...well...then present something else than PR.

That statment goes for you as well, you can type anything you want, but your point is clear, you're not educating anyone trying to explain, you're making your self try to look credible and gathering attention, with your worthless posts. I don't care about karma. There is a fine line between critique and total nonsense.

 

[Lonbjerg]

That statment goes for you as well, you can type anything you want, but your point is clear, you're not educating anyone trying to explain, you're making your self try to look credible and gathering attention, with your worthless posts. I don't care about karma. There is a fine line between critique and total nonsense.

You are getting personal now, no need for that.
All you need to do is stop the PR and present some testable code.
It's not about me...it's about the MIA Mantle...remember?

 

[Carfax83]

And how do you know that?

Eurogamer did a very detailed face off with the next gen consoles including frame rates

I don't understand why some of you are fighting/resisting Mantle, if Mantle lives up to expectations, it'll move the industry forward, you'll get a better experience out of games, maybe you'll start seeing new types of games that you've never seen before, etc...

If it fails, then Direct3D will still be there, OpenGL will still be there, so for the "non believers" Mantle living up to expectations can only bring good things, if it fails you'll lose absolutely nothing, so I don't understand why you guys are resisting this, it has the makings of something very special.

As I, and several others have mentioned in this thread, Mantle comes with some severe penalties, most notably the sacrifice of backward compatibility (both software and hardware), as well as cross architectural compatibility.

Those are very good reasons to "resist" Mantle imo..

 

[desprado]

All the facts in this thread from both sides are all purely opinions from people who basically have zero knowledge of game design, and how games utilize CPUs and GPUs.

There are a couple people on this forum who I believe have some good knowledge of the topics at hand. They are not participating in this thread however. If they are then I guess I have been hoodwinked because of all the ill formed and nonsensical posts bordering on complete irrelevance I have read in the last three pages.

I also don't care if people have negative things to say about mantle in the thread. I do take issue with the fact that they make no sense and have no relevance to the mantle thread.

You'll all notice how I'm quiet in the nvidia game works thread. It is because I try desperately to live by a simple rule which is if you do not have anything good to say do not say anything at all. I try and stick to that but the trolls in this thread are ruining one of the few interesting threads active in the forum...

What u want to them to talk.There is no benchmark,demo or any release date expect PR Slides and Gaming evolved partners interview.