Maxwell or Mantle makes you more excited?

Considering that slide, I think more people will buy Kaveri systems this time around. And with the possibility of problem free Dual Graphics means that Kaveri + Bonaire GPUs should work very well in Mantle games. In mid-range notebooks especially AMD systems will have big advantages against both Intel + AMD and Intel + nVidia systems. As far as mid-range systems go, even without mantle, the CPU side of Kaveri will not be a bottleneck. Sure quad core intel still has a lot of advantages in productivity applications but hopefully that will change in favor of Kaveri once (hoping it does) HSA and hUMA support takes off. As for the high end lets hope MSI can convince AMD to allow them to change the APU TDP to 45W.

 

Now that Mantle has been confirmed to not be locked to AMD cards only can I change my vote?

 

What a total BS, mantle won't make games run at 1200FPS and as far tearing goes that image is also BS (either faked or run without mantle). Maybe it happens more on nVidia cards or its a crossfire or sli issue, but i haven't seen it happen on my system yet, and why would you bother going above 60 fps when aiming for minimums around 30fps is the right thing to do. Seriously more people need to run benchmarks the way [H]ard|OCP does.

Edit: I just looked at both images at separate sides and it has totally been faked. First they enlarged the picture to avoid the edges because it has been cut into vertical sections and re-positioned, then they also shaved off the top and bottom to make it seem like a similar sized image, yet the aspect ratio isn't even the same (the first image is clearly taller than the second, they didn't shave enough). Nice try to whoever did it but at the end, it's EPIC FAIL!

I can tell that you have a nVidia GPU as your avatar, and can see where the bias is coming from, but please, before posting stuff like that make sure its legitimate.

 

Where did you read this? I have only heard Dice say they want it to be open. I have not heard AND say it will be open, nor have I heard Intel or Nvidia say they will use it. And open is different than a license.

 

AMD sheds more light on Mantle

AMD’s Mantle Does NOT Require GPUs With GCN Architecture | DSOGaming | The Dark Side Of Gaming

http://www.redgamingtech.com/amd-ma...100000-draw-calls-similar-to-ps4-api-library/

 

To make it more clear, AMD themselves have stated that Mantle will be open. The keyword here is 'will' as currently it is not open. AMD said they will make it open in 2014, and even gave a specific date I believe (or the month at least), after that anyone, including mobile gpu makers (Adreno/Qualcomm, Mali/ARM, Imagination tech are their HSA Foundation partners after all along with chipmakers like Samsung) can optimize the API for their GPUs. I don't have reference links right now but this info came straight from AMD.

Edit: Here it is http://vr-zone.com/articles/mantling-alliances-ritchie-corpus-amd-interview/58215.html
VR-ZONE interviewed AMD's Ritche Corpus and he said that they will make it open once the software is stable, so I might have been wrong about the exact date or I found it from another source.

 

Looks like the obvious went right over your head.

 

Wow, more people here are actually leaning more towards a promise about ponies riding on a rainbow, than a proven shift toward better architecture which will 100% give way better performance than whatever a piece of software can do. Mantle still havent been showcased or proven, other than good words from AMDs partners.

GG notebookreview.


And incase someone missed it earlier

John Carmack: Nvidia's OpenGL extensions and AMD's Mantle offer similar improvements over DirectX | GearNuke

 

Wait, you have Maxwell benchmarks already?

 

Im not saying Mantle wont have any benefits. It will most likely do, but Im having hard time picturing a new API beating DX which have been here since 95 and improved year after year, by 20%. That is almost the same improvements as the jump from GTX 580 > GTX 680 saw.

If DX is so bad, why havent anyone done anything about it earlier?

No but you got to be pretty blind to not understand that Maxwell will be a decent upgrade from Kepler. I havent said anything about percentages, but it wont be low.

 

Because it's Microsoft...

 

But its not microsoft thats using it, its the game developers right?

 

Well that was just disingenuous and you weren't helping your argument for G-Sync by using that BS image.

 

No it really wasn't. The point was that no matter what FPS boost Mantle provides (so far only in BF4..at least if AMD is to be believed), it doesn't address a fundamental flaw with current display technology. VSYNC and other implementations currently available are poor substitutes for what G-SYNC brings to the table. What is disingenuous is claiming that screen tearing and vsync input lag isn't a problem. Right now AMD has absolutely nothing to counter that. And the suggestions that Mantle is in any way compatible with NVIDIA hardware is laughable. It was built with GCN in mind, AMD can claim it's open all it wants but it's not. And there's zero chance NVIDIA would ever use it so it will be relegated to a few select EA titles. Since Frostbite is limited to EA only and going by how crappy BF4 has been (netcode, crashes, graphical glitches), things don't look so peachy. Like Cloudfire mentioned, if it ever came down to it, OpenGL offers similar gains for NVIDIA hardware with respect to draw calls and has been around for years and works on Linux and Windows.

Maxwell + G-Sync is a far more potent combination than EA's poorly coded games + Mantle. Oh and show me some hard numbers of Mantle gains? Because so far all its riding on is a bunch of hype. G-SYNC has credible developers (including Johan Andersson) and web editors gushing over it:

P.S. Who in their right mind trusts AMD with any software implementation? Their track record is one of the worst in the industry, especially with Crossfire. They finally introduced frame latency/runt frame fixes in Hawaii only after a community uproar and learning that NVIDIA had been doing it (and still does it better) at the hardware level for years yet it is still not fixed for R9 280X and below. Don't even get me started on their poor drivers which I've had years of experience with, especially in Crossfire and even more so on the mobile front.

 

You're just an Nvidia shill and nobody takes anything you say seriously since you like to spin numbers and make up stuff thinking it will your arguments.

D3D has many shortcomings but M$ hasn't done anything about it because why would you when you have a monopoly. I hope Mantle becomes a wakeup call for both M$ and Khronos and influences the future direction of their respective API's.

 

Because AMD wasn't powering both the Xbox and Playstation earlier?

What about 3dfx?

AKA the same type of reasoning people who are optimistic about Mantle are using.

 

So? How do you know Mantle + G-Sync (or a similar technology) won't be a thing in the future? If you'd stop fanboying and shilling for a moment you'd realize that both Mantle and G-Sync are amazing and PC gamers deserve no less than to have both. And as for Johan Andersson gushing about G-Sync, what do you expect him to say at an Nvidia event he went to on Nvidia's dime? Just saying. Not that I don't think G-Sync won't be game-changing. It certainly has the potential to be, but so does Mantle, in different areas. They don't clash at all and would be perfect complements. And you'd be blind to think it's all roses for G-Sync from this point on. It will have major hurdles with adoption just like Mantle will. G-Sync is not out in people's hands yet and in fact Mantle will beat it to the punch with the BF4 update next the month. So to use your own words, all G-Sync is riding on is a bunch of hype, just like Mantle is, but Mantle will be able to strut its stuff sooner.

And if Nvidia decides not to support Mantle in their future products, that's their problem. There's nothing preventing them from doing so since Mantle is not tied to GCN. But knowing that Nvidia apparently loves to hurt the industry and PC gamers by sticking to closed and proprietary systems (see PhysX, CUDA, crappy Linux drivers), there certainly is a precedent for it.

 

And where's your crystal ball predicting Mantle will take off? Given the reduced role AMD has in the industry and it's track record, the chances are very small. So far their biggest supporter is EA with its proprietary Frostbuggy engine.

So now your hero Johan Andersson is a paid shill too? I guess everyone that doesn't agree with you or has something positive to say about NVIDIA must be a paid shill. What a joke.

Well it's good to know you think G-SYNC will be a game changer..laugh. Mantle won't get any traction unless NVIDIA, who has the lions share of the discrete and mobile GPU market, adopts it and good luck to AMD with that one.

Let's use some logic here: API that is limited to EA's rushed Frostbuggy games and AMD's GCN (only the gullible believe otherwise) vs G-SYNC that will be ubiquitous to nearly all games. G-SYNC is restricted to NVIDIA but has no software limitations like Mantle and can gain wider adoption much easier and faster.

No it's Mantle and AMD's problem because it will be dead before it even gets going.

Yeah your credibility is zilch after that statement. When Mantle fails, I'll bookmark this thread and remind you of it.

 

These are all statements from Dice saying it will be open, or not tied to AMD. I still have not seen anything official from AMD, nor details on how it will be done. Will it be a license, will it be totally free, can Nvidia or Intel modify the spec, etc? Until then it might as well be tied to AMD. After all I don't see AMD licensing Physx, which also is not "locked" to Nvidia hardware, as it could run on AMD hardware, or even Intel hardware better than it does now.

 

Wrong, wrong, and wrong. Besides the 15 games powered by Frostbite, Mantle will also be in Star Citizen, in Oxide Games' next-gen Nitrous Engine, and in the future titles of Square Enix's Eidos-Montréal studio such as the upcoming Thief. And you're conveniently forgetting that G-Sync requires special hardware, namely the proprietary Nvidia board which replaces the monitor's scaler. And the G-Sync scaler will only be in a limited selection of overpriced gaming monitors with TN panels.

And so is yours. Difference is, I'm a nobody but you write for a couple tech review websites so people expect at least some degree of objectivity from the stuff you write. Your credibility is definitely called into question when you're clearly biased toward Nvidia as has been the case here, and insulting your readership when they're critical of your flaws doesn't help either.

 

We don't know which monitors in the future will support g-sync, or even if notebooks will. It is too early to say. All we know is the current monitor that will support it.

 

Have you thought that maybe this is by design? Maybe distancing Mantle from AMD is a way to make to make it more enticing to other platforms and vendors. After all, the dream of Mantle is cross-platform and cross-vendor adoption. I say be patient and there will be a lot more information in the near future. Sheesh, I don't see you getting this worked up about G-Sync.

And no, this is not like PhysX. PhysX is cleary proprietary and needs a CUDA-enabled GeForce GPU to run at its best. It was originally developed by Ageia and then bought by Nvidia. And why would AMD pay money to use crappy PhysX when they can use DirectCompute or OpenCL for game physics simulation for free and allow it to run on any GPU, like they did for TressFX in Tomb Raider? This is the point I was making earlier about wishing that Nvidia would take the high road and adopt open standards instead of becoming more and more closed. It's the best thing for everyone.

Exactly. G-Sync clearly requires its own ecosystem, and how wide the ecosystem becomes remains to be seen. As of right now, it's very exclusive. G-Sync is only a thing for desktop owners with Kepler GPU's and special monitors. All of us here won't be able to use G-Sync on our notebooks, and of course AMD and Intel GPU owners are left out too.

 

I would like to see both efficiency improvement to get better framerate, quality at lower power consumption. 30-50% reduced power consumption to produce similar capabilty to ivy/haswell line-up accordingly and 30-50% better efficiency for "mantled" gaming software compared to if mantle optimization is not implemented.

All in all, I'd love to see the efficiency to increase by 60-100% next year by hardware and software optimization. :thumbsup: Probably, game with mantle first in order to better utilize and prolong my current notebook gaming capability, hence no need to replace it next year

 

Unless AMD can convince Epic, Crytek and other engine developers that make up the bulk of PC gaming (and now making inroads into consoles), Mantle will be a niche API that depends on AMD paying off software devs to use it. The 15 games from EA you listed is proof of that fact plus Eidos Montreal is a paid Gaming Evolved partner. And if NVIDIA is compelled to create a competing API, then AMD will have helped fragment PC gaming even more.

With regards to G-SYNC, everyone is aware it requires g-sync circuitry to operate, that shouldn't need to be stated. However, the fact that it runs on most games without additional coding means that uptake will be fairly quick. As for the TN monitor argument, companies like Overlord are already interested in the tech and there is no indication the technology is confined to TN--if you've got proof otherwise I'd like to see it. On that subject, for gaming purposes nothing beats TN. Most IPS displays on the market have fairly low end panels so the vast majority you see in stores are mediocre junk that are easily matched by these top end TN panels. I have a beautiful $1000 wide gamut IPS panel that I'd give up for a 27" high end TN with g-sync without hesitation. Why? Because the tearing + blurring is all to obvious in FPS games.

My opinion on forums is my own and not reflective of any tech sites. In a perfect world, I'd love to see G-Sync and a form of a Mantle-like API used by AMD/NVIDIA and all game devs. Realistically that won't ever happen.

P.S. Someone should put up a poll here on NBR: Would they take a mantle equipped gaming notebook with a IPS panel or a maxwell one with a high end TN panel + gsync? There's nothing stopping a vendor like Alienware from integrating g-sync into their notebooks. Don't be shocked if you see it in a Clevo/AW high end notebook in the future.

 

Actually I am pretty worked up about g-sync also. My big question is why we have not seen independant reviews on it? Why haven't they let some big sites in to just play with it using various games? Maybe they still need to tweak the hardware or software?

Any way, all of these are unknown, Maxwell, G-sync, and Mantle.

 

Unreal Engine might be a hard sell considering Epic's historic relationship with Nvidia. I'm not sure about CryTek since Mantle support in Star Citizen, the biggest CryEngine game to date, has already been confirmed.

As for the poll, you'd have to convince people of the benefits of G-Sync without them having seen it in-person because G-Sync is one of those things that needs to be seen to be believed. On the other hand, the benefits of Mantle are easier to explain and grasp. I predict the results would be quite close because some people just aren't bothered as much by tearing, or they run all their games with V-Sync anyway, and would rather have 20-50% performance improvement plus the much nicer IPS panel. I would go for Mantle and the IPS screen, but if it was a straight-up Mantle vs. G-Sync poll, I wouldn't vote because I want both. It's unfair that we as PC gamers have to make these kinds of decisions and this needs to change.

 

Pretty heated debate. For someone who is not intending to upgrade anytime soon the fact that mantle will work with previous gen GCN such as my 7970m is great. I don't really care what the difference is, its just good they are still supporting my old card. Hopefully better driver support too.

Maxwell top end cards will cost in excess of $1k? Did Titan push the limits of GPU pricing up forever? No thanks.

So will g-sync only work on monitors that support it? That's no better an mantle only being supported by EA. Stupid industry.

 

What I desire more than anything are games, especially AAA expensive games, that launch without constant server crashes, without a locked pre-selected server farm, without significant bugs and glitches. Mantle and G-sync are gravy on the cake as far as I'm concerned. Neither mean squat if these major game developers can't get their act together.

 

"Paid shill" - No I`m not paid by Nvidia. I`m critical to anything which haven`t been proven yet. Especially when it involves AMD aka "Bulldozer will be so great" and AMD partners boasting a technology which AMD have paid them to use. If you don`t see the dangers here, god speed.

"Nobody takes what you say serious" - Take a look at my post count. Take a look at my rep points. Take a look at yours. You reply back to me. Seems like someone is taking me serious. But I`m not taking you serious. You are just an insecure little kid throwing around personal attacks because someone dared to be sceptical against this magical API.

"Microsoft has monopoly"
Why do you think Mantle was made in the first place? If a simple software API can increase performance by that much, why haven`t, lets say Nvidia made it earlier, or AMD? If 20% increase in performance were true, heck why waste billions of dollars on making a new GPU, R&D cost, new fab equipment and such when you can just make a piece of software? GTX 580 > GTX 680 saw a 20-25% increase. Are you telling me a brand new API will offer the same?
Then back to my question: Why haven`t anyone done anything about it earlier? Game developers use DX everyday. They should know everything about it and be the first one to complain if overhead and such get in the way with DX. They haven`t complained. Because DX might not be as bad as some people make it out to be.

Are you actually serious? You are saying that hardware development have never happend before, Fermi never existed, Maxwell might bust just like Mantle could do?
I guess the idea of putting on more cores on a GPU could all be for nothing then.

That argument was weak r3d. Everyone knows that better hardware give better performance. And better hardware comes out all the time. New API on the other hand. That is something no one here have any grasp of what the effect will be.
Just sheeps, and hopeful people, eating everything AMD tells you raw.

 

Yeah funny how they always tell people they are paid shills when they do not agree with you. Lack of arguments that is replaced by personal attacks. How sad.
Did you notice that he was quick to call you out on your credibility just because you chose to be sceptical about something thats never been proven?

Its not like your participation in various forums for several years, or your benching, matters right?
Best to ignore these people.

AMD will have to cough up a lot of money to make developers code their games for AMD GPUs.
If Mantle becomes universal and open, Nvidia will not use it. They are concentrating on making OpenGL improvements and according to multiple sites, that will offer the same improvements as Mantle will do. So AMD will still be alone with their own API, which they will be forced to pay developers to use, since DX will be the choice of software. And OpenGL which already exist.

Tim Sweeney from Epic Games explained why OpenGL will have much greater support from game developers than a new API like Mantle,

AnandTech | Tim Sweeney, Johan Andersson & John Carmack @ NV Montreal - Live Blog



Also, Mantle doesn`t seem to scare Microsoft.

AnandTech Portal | Microsoft Comments on State of Direct3D, Mantle Support, & Names XB1 API



So to recap:
OpenGL will offer the same improvements as Mantle. Nvidia is working along with Valve to make optimizations there for the SteamBox. And the fruit from that work will be used on Nvidia GPUs.
Microsoft is still improving DirectX and will be using it on Xbox One, and it offer low level optimizations, to make most out of the hardware. Will be used by the clear majority of PC game developers.


So yeah. Maxwell + G-sync (which have unlike Mantle, been showcased and proven to work) seems to be the way to go. Mantle will be no threat, and both software optimizations and hardware development will go its usual course :thumbsup:

 

AMD is not Intel, they don't have the money to pay people to use their stuff. It's just that is nVidia is one of those companies that are hard to work with, so increasing number of developers turn their attention towards AMD. The frustration of Linus Torvalds towards nVidia is the best example.

 

How is AMD suppose to get developers to use Mantle, if they have very little funds to give the devs? DX is priority since GCN (7000 series) is the only GPUs that can use Mantle. Nvidia won`t support Mantle when they have OpenGL. Intel will surely not support Mantle with their IGPs, since AMD is the competitor. Coding with Mantle too will require more work hours and somone got to pay for that right?

I`m not sure the frustration toward Nvidia is so big that they will prioritze Mantle instead of DX. Youre right though, Linus is not actually happy with the crappy Linux support.

 

AMD isn't trying to get developers to use Mantle, I think it was Johan Anderson who came up with the idea so he was went to GPU makers supposedly (including Intel) to get support for it. Apparently Intel and nVidia turned him down but AMD's Matt Skynner loved the idea of having a close to metal API to boost graphics performance (he is the head of AMD's GPU division after all) and called it Mantle, which is the next layer after the core (same for a planet and the gpu(sometimes referred to as metal), since the core is the basis of both). Now it is still not tied to the core so other manufacturers (I think AMD made it open because of their partners in the HSA foundation) can use it after putting in the resources to optimize it for their architecture (it costs money to optimize, hence AMD's older GPUs, including VLIW4 in Richland will not support it). Now since this was meant for the Frostbite engine, 15+ games will support it in the near future. The rest of the developers just jumped on the bandwagon. For example Chris Roberts already said before that 2013 GPUs won't be good enough to run Star Citizen maxed out, so it was a no-brainer for him to adopt it. Developers of thief were already tied to AMD because of TrueAudio so there was no reason for them not to adapt it.
Don't get me wrong, Mantle is not designed to replace DX11, or OpenGL for that matter, its just an alternative to DX11, one that covers its shortcomings, mainly cross-platform support (yes, it can be adapted by mobile gpus, so that developers don't have to code for too many operating systems and architectures separately), the fact that PS4 code is very similar to Mantle is already a win, but Mantle will still be a secondary API on Windows based machines next to DX11, and in fact will not affect DX11 coding much, it will just be a side-project for porting from PS4 in the near future except for PC exclusives. OpenGL on the other hand, even though its somewhat multiplatform, is still much harder to code for than Mantle, so I guess Mantle will be more of a competitor to OpenGL.
They said that coding for Mantle only takes two Man * months, so a good dev team of a dozen people will get through it in a week or two. I don't think it will do much of a dent on development costs and profits can only grow.
The frustration was mainly on the Linux side where DX11 doesn't exist but it might have been a bit of a pain on the Windows side too, so devs crossed the border. This usually equates to better AMD optimization in game at release, but again this is for DX11. As far as I can tell people code for DX11 first, but developing on AMD hardware might be the catalyst to get devs interested in coding for mantle too. Now all of the above was assuming these were games based on multi-game engines developed by big studios. One can only guess what indi developers will do.

 

AMD doesn't need to pay devs to use Mantle. Mantle has enough merit that developers of cutting-edge games will support it of their own accord if it means they can squeeze more performance out of high-end PC hardware and create richer experiences. Developers like Chris Roberts:

And I don't know why everyone is so high on OpenGL right now. Even John Carmack will tell you that Direct3D is the better API today. OpenGL is severely marginalized and has zero market penetration in AAA PC games. Not to mention that even though OpenGL is standardized across platforms, it's not the same across platforms. There are significant differences in feature set between OpenGL on PC's and OpenGL ES on mobiles and WebGL and the PS3 API and on and on. At least Direct3D doesn't suffer from this problem, but Direct3D is monolithic and has way too much inefficiency and overhead.

And like Link4 said, Mantle doesn't take a lot of time and work to code for. Oxide Games quoted 2 man-months of work. That's a month with two people and two weeks with four people. Devs could easily release a game with a D3D render path initially and then patch in Mantle support a few weeks later. It's not a problem. This is the approach DICE has taken with BF4.

Oxide was demoing their StarSwarm game using Mantle at APU13 here are some tweets from the event:

It doesn't take a genius to see that Mantle is a ridiculous improvement over Direct3D and OpenGL.

 

This page has been pretty informative because of all the links people have posted to prove their points. I've learned quite a bit about Mantle and G-Sync (although G-Sync wasn't really the topic). It is unfortunate the same posts that contain these links will probably lead to a thread lock.

 

I don't think you quite understand what the word "proof" means.

The reason why AMD is doing it now is because having GCN on the consoles puts the in a position where devs can just easily port over the low level code.

3dfx did. The average performance increase was >20%. Man, all these people who think that Mantle can do the same thing must be delusional, right?

People have been complaining about the draw call issue years before Mantle was even announced.

Wut. Here's your original post:

I have no idea how you could claim Maxwell will "100% give way better performance" than Mantle given that you have no performance benchmarks for either of them. Somehow this turned into me saying Fermi never existed.

 

I believe AMD called it Mantle to tie-in with their Volcanic Islands series GPU's.

 

That's the other interpretation people came up with, just like people interpreting the name Volcanic Islands for GPUs because they are running too hot, but probably it was a combination of both. If tying with Volcanic islands was the reason I think a better name would have been Magma, but who knows. My assumption was based on the fact that AMD showed all those slides with the planet layers way too many times. What do you guys think about this, I find this topic quite entertaining, makes you relax too by getting your off of complicated stuff.

 

Yeah I know the exact diagram you're referring to, and I've also seen it one too many times. I think all of these are perfectly valid explanations!

 

Intel might won't support Mantle with their IGP, but they certainly not against if the games will better utilize their CPU cores. They do not care if Nvidia or AMD GPU near their CPU...

Ohh, thanks for reminding me for 3Dfx Glide, that was not just faster API than DX7, but also lot nicer! Too bad Nvidia killed it!


By the way, already preparing myself for a Mantle test with a mid-range mobile GCN Radeon, hope it will perform as expected;

 

Atom Ant and R3d was talking about HSA without PCIe getting in the way. NVidia only makes discrete GPUs.


On the other side, game devs are generally more happy with Intel's CPU offer.

 

Let us see, but i'm kinda septic about this mantle thing..

 

You're septic? Ewww.

 

I'm too, maybe it is just like the OpenCL acceleration staff, what we have heard about two years ago and I still do not know program which would take advantage of it.

 

It's currently used a lot more in GPGPU and professional applications than in games, e.g. Adobe Premiere Pro. Hopefully that will change.

 

I voted for Maxwell, but I'm probably more interested in what AMD brings to the table for 2014. It's time for them to step up to the plate.

 

They don't talk about it on the Stream front page doesn't mean the don't do it. Plenty of games already do.

Some old examples (with CUDA).

With HSA, the iGPU compute throughput will also be useful in high frequency physics.

And here's are some voxel / engine / demos. The whole world is generated and rendered, ON THE FLY, with a Radeon 4770.

 

I have a feeling that the future of Mantle is tied to SteamOS (and vice versa)...

If Valve decides to REALLY screw MS, they might decide to go the Mantle way (quite convenient with all next gen consoles running on AMD hardware) to gain a competitive advantage over MS DXxx.x

Wouldn't you be happy changing MS's schizofrenic Win8.X AND in parrallel get a 20% boost in performance??? To me sounds like a killer deal, but let's see how the whole thing will evolve. Exciting times ahead, no?

 

Another game coming with Mantle support. Sniper Elite 3 to tap AMD's Mantle API - The Tech Report
Looks like we will get even more next year.