Wouldn't developers absolutely hate this? Mantle is already supposedly a lot more work for devs from what I've read.
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
CES: Mantle API news thread [BF4 up to 45% faster, Star Swarm demo flies on Kaveri]
- Thread starter DieH@rd
- Start date
If it becomes a choice scenario- I'm pretty sure Mantle will be the one of choice for strategy choices due to Oxide Engine, but most likely I think we'll see games shipped to the lowest denominator if it really becomes a major issue.
Felix Lighter
Member
Stability-wise, anecdotally, for myself and those I play with, it seems pretty solid recently. As for gameplay bugs and balance issues, their plate is more than full but hopefully the engine devs are less involved in those problems.
Osiris
I permanently banned my 6 year old daughter from using the PS4 for mistakenly sending grief reports as it's too hard to watch or talk to her
Yes, that would be as much the goal as competing with a technological solution directly.
Make it more work, cause both initiatives to fail, fall back to DirectX, rely on/push Microsoft to respond with their own low-level DirectX-like vendor-independent implementation. (I have heard MS is already on this path).
Microsoft will never allow DirectX on other platforms.
edit:
Let's see benchmarks in actual games first. Up to coming from producer isn't good indication. Though Oxide comments and Dice comments on CPU aspect of Mantle is very good. Generaly what they say is that games with Mantle won't need good CPU. So you won't need to buy i5 to play new games, you will be able to play games on something like i3 or AMD 4 core.
Aren't most of BF4's problems server-related? In that case the rendering technology is separate from those bugs, especially now that the Frostbite guys became their own thing.
I'm just going by what I read though. I personally don't care about BF4.
Osiris
I permanently banned my 6 year old daughter from using the PS4 for mistakenly sending grief reports as it's too hard to watch or talk to her
I meant cross-platform as in multiple GPU vendors, rather than the single vendor situation of Mantle, I have changed it to vendor-independent instead in my post.
misread sorry/
So it seems the entire premise of Mantle is to get rid of or minimize the CPU overhead inherent in some DX driver calls, coupled with much smarter CPU threading when dealing with modern rendering pipelines, right?
If that's the case then I would have to agree that the greatest gains will be on games that not only have demanding graphics, but ALSO demand a lot of the CPU. Things like strategy games, games with complex simulations running under the hood, games with lots of interactive/reactive phyics effects. Imagine what that would mean even for low/mid range specs- actually SPECIALLY for the low/mid range gaming PCs! Even non-high end PC's have CPU's multiple times more powerful than what's on next gen consoles. Freeing them from the burden of being slaves to the renderer could mean big things for lots of different types of games on PC.
I wonder if Nvidia's answer to this will be to implement Mantle on ti's hardware, or will the rumor of embedded ARM processors to handle all that CPU overhead turn out to be true? An embedded ARM chip, IF it could handle all o fthe driver overhead would be an even better solution as it would work out fo the box with other OpenGL/DX games. They might not be able to take advantage of other special features without reworking their code, but right out of the box the CPU rendering overhead would be gone.
If that's the case then I would have to agree that the greatest gains will be on games that not only have demanding graphics, but ALSO demand a lot of the CPU. Things like strategy games, games with complex simulations running under the hood, games with lots of interactive/reactive phyics effects. Imagine what that would mean even for low/mid range specs- actually SPECIALLY for the low/mid range gaming PCs! Even non-high end PC's have CPU's multiple times more powerful than what's on next gen consoles. Freeing them from the burden of being slaves to the renderer could mean big things for lots of different types of games on PC.
I wonder if Nvidia's answer to this will be to implement Mantle on ti's hardware, or will the rumor of embedded ARM processors to handle all that CPU overhead turn out to be true? An embedded ARM chip, IF it could handle all o fthe driver overhead would be an even better solution as it would work out fo the box with other OpenGL/DX games. They might not be able to take advantage of other special features without reworking their code, but right out of the box the CPU rendering overhead would be gone.
longdi
Banned
considering that Intel will be selling a 8 cores haswell for $600 later this year, it makes sense AMD is trying to minimize cpu reliance for PC gaming. I think its ok because i have an AMD GPU, but i dont expect more than 10-15% better performance with Mantle if you already have a powerful cpu
jacksepticeye
Member
But I just got a 780Ti 
Then again more games will need to support mantle won't they?
Then again more games will need to support mantle won't they?
I doubt very much they are flat out lying. Imagine the outcry when people eventually test it out.
45% is probably in the ideal situation but i don't think they would lie.
emiranda061288
Banned
This just means that certain multicore cpus are used more efficiently, up tp 45%
dr_rus
Member
You should actually expect worse. These figures are for a weak CPU, dealing with inefficiencies of which is one of the better points of low lovel api compared to high level api. On a good CPU you'll probably see much less than 'up to 45%'. This probably is also the main reason we don't have any high end mantle benchmarks yet.Dont forget people, those are numbers for Kaveri APU. Im hoping for even better numbers for R9 290/290X GPUs.
Mantle allows for the GPU to handle some otherwise CPU-bound tasks. So if you have an absolutely crippled CPU bound to a decently powerful GPU in a terribly unbalanced situation and suitable code that overly relies on the CPU, adding in Mantle to move some of those CPU tasks to the GPU would provide significant savings.
In the real-world, people tend to have more balanced systems and games tend to be coded for such configurations, so Mantle will make less of a difference.
Zombie James
Banned
Damn, 45% is impressive. Makes DirectX look like a fucking anchor.
Has MS showed any interest in the PC gaming market lately other than forcing us to buy the newest Windows for the newest DirectX?
DiddyInAHellcat
Member
This sounds too good to be true. Hopefully it is tho
KojiKnight
Member
Make no mistake... Nvidia likely won't go Mantle, but likewise they aren't likely to make a competing API. Instead, what they'll do is continue to double down on their own features. DirectX won't be disappearing even if Mantle becomes a norm in a lot of games. Not just because of Nvidia, but because there are a lot of legacy devices that still run games that they won't want to alienate.
Mantle will become AMD's physx. Games will still be mostly DX with some vendors supporting primarily AMD's mantle or Nvidia's other features... just like they always have and always will.
Mantle will become AMD's physx. Games will still be mostly DX with some vendors supporting primarily AMD's mantle or Nvidia's other features... just like they always have and always will.
Turin Turambar
Member
Yep. The first thing one can think when hears there is a new API for gpu is "cool, now my gpu will be as if it was more powerful!".
But most of the benefit isn't in the gpu, but in the cpu. Much less overhead for each draw call in the cpu and linear scaling as you increase cpu cores, but Mantle isn't going to magically increase the number of polygons or texels or shaders per second it can do your gpu.
Of course, in a indirect way it can increase the final framerate: if your cpu waste less time per cycle thanks to Mantle, it can do more cycles per second, and increase the fps. But that will be noticeable in cpu bound scenarios mainly.
KojiKnight
Member
A note to the people who have super high end PCs and CPUs complaining you won't likely see as much improvement... I think you're missing a key point. The reason why your CPUs aren't being pushed as hard is because if they were, mid grade PCs will not be able to keep up. If mantle DOES become a standard, and middle ground PCs have their CPU freed up, what do you think that means for future games?
More games will then take up the slack of extra CPU time, and thus your high end machines will also gain performance.
More games will then take up the slack of extra CPU time, and thus your high end machines will also gain performance.
I went back to nvidia because well G-sync and Shadowplay for example. I had a 7970 very good card but don't know i always had nvidia easier for downsampling and Nvidia inspector to tweak games even further etc. There on the good way with AMD..competition is good but G-sync floored me the moment it was announced.
Mantle doesn't do anything of the sort.
Preformance acceleration of old games is a side product of slimmer and more optimized API. We need to see how much preformance gain will Thief get in late February.
More importantly, Mantle will enable creation of games that were not possible on DX. We shall see how will PC-centric games [real time strategy genre and space shooter especially] take advantage from this tech. Those games will have even more preformance gain from DX.
Coolbgdog12
Member
I have a 2GB 7870 and a FX 4100. BF4 plays pretty good on some maps online while some others drop to 30 to 35 fps periodically. Mantle would theoretically boost my fps on those maps from 35 to maybe 50-60fps at all times?? Am I dreaming?
The whole Kaveri line and Mantle seem to be the right step for entry-level gaming where people are more casual and don't care about pumping up the notch on their settings.
My next laptop will be the flagship chip Kaveri system, whenever it's available. I only play a few PC games, but the ones I do would definitely be better on this platform. It's a shame intel didn't push it's Iris Pro 5200 line to laptops with a 17" screen, or I would have gone that rout.
My next laptop will be the flagship chip Kaveri system, whenever it's available. I only play a few PC games, but the ones I do would definitely be better on this platform. It's a shame intel didn't push it's Iris Pro 5200 line to laptops with a 17" screen, or I would have gone that rout.
That entirely depends on if the game is CPU demanding on let's say 64 player conquest
Yeah that will only happen if Nvidia and preferably Intel Graphics will support mantle. Or do you think Devs will design games that will not even target half of the pc market? Honestly if we ever see much in that direction hinges on whether the competition adapts API. If that doesn't happen they will still have to create a DX renderer and design the so it works well with DX.
ScepticMatt
Member
Sorry I misread the slide as Mantle enabled games.
So it seems that high end desktop benchmarks are still to come?
So it seems that high end desktop benchmarks are still to come?
Dictator93
Member
They could just use the newer GL extensions which give similar CPU gains for draw calls.
Dictator93
Member
I will still laugh at it given how it is pretty often referenced in combination with that horribly non-topical JC quote and a huge exaggeration.
Of course, PC gaming going away from Microsoft trying to create a walled garden is a great thing.
EatChildren
Currently polling second in Australia's federal election (first in the Gold Coast), this feral may one day be your Bogan King.
I don't see Mantle having any impact on game design itself at least in the near and foreseeable future. It's an incredibly optimistic outlook that requires Mantle success explodes and AMD GPUs become a standard for PC users, so much so that developers can comfortably bring API exclusive titles to the PC environment without fear of market fragmentation and loss of sales from those who bet on team green.
Well, one thing going for mantle is that it is similar to PS4's API, and some, if not all optimizations catered for one, can be used on both. This was mentioned in slide 31 in http://www.slideshare.net/DICEStudio/mantle-for-developers .
So devs partnering with AMD may be inclined to use optimization they did for PS4 on Mantle.
OldAsUrSock
Banned
Mantle easily because g-sync requires two purchases while mantle requires one.
That might have been becaused when you suggested was a time when they said GCN only.
BF4 is CPU demanding. I just want random drops to go away. 48 player match running 100fps then 50fps gets slashed off randomly. I would like to run with better graphics, but the random drops get worse.
googleplex
Member
I was already leaning towards using ATI over Nvida for the rig I'll be building this year. I think this may have cemented that decision.
It doesn't solve the problem with a 60hz monitor with v-sync on/triple buffering screentearing will persist when i think of it. And it doesnt do a variable framerate like G-sync does with monitor and videocard.
Mantle itself doesn't need to have an affect on game design anytime soon.
AMD's GCN hardware is already impacting game design in a major way. Every large studio making multiplatform games is going to be familiar with a somewhat powerful implementation of the GCN architecture in PS4 and X1.
PC and PS4 are most likely to be the lead development platforms for multiplat games going forward. All that's important right now is that games start coming out this year with Mantle support, and that Mantle provides a decent performance boost to users of GCN hardware.
I would assume that AMD is smart enough to make the console API's as similar to Mantle as possible to capitalize on the hundreds of new developers working closely with AMD hardware and API's at a console level. It's possible that there will be rapid improvements to Mantle and(hopefully) AMD's driver scene period.
DirectX came out in 1995, so we're talking 20 years ago, not 10. Also "every GPU manufacturer had its own low-level API" is false. 3dfx had GLide, but that's it. Proprietary OpenGL extensions was the closest we got to card-specific APIs.
You have to go back to the DOS era to when PCs were coded to the metal. And it fucking sucked. Hey you just spent a few hundred bucks on a nice soundcard? Well only half of your games support it, enjoy!
DirectX was shit in its early years. Its started catching up steam in 1997-8. Arrival of first Gforce sealed the deal.
IAmRandom31
Banned
As I wrote in the other thread ...
Mantle + SteamOS will allow for near console style programming for developers. Low memory overhead from the OS, really nothing getting in the way, and much lower level programming capabilities.
Not surprising at all to see this level of a boost working with the mantle API.
DirectX will be pretty much gone within 4-5 years I think.
Mantle + SteamOS will allow for near console style programming for developers. Low memory overhead from the OS, really nothing getting in the way, and much lower level programming capabilities.
Not surprising at all to see this level of a boost working with the mantle API.
DirectX will be pretty much gone within 4-5 years I think.
Mantle alone no. But GCN in every next gen console + PC AMD market share can seriously change things a bit.
imo this is most interesting part of earlier presentation:
cpu parallelism up to 11
s_mirage
Member
Nope.
3dfx had Glide
Rendition had Speedy3D and RRedline
S3 had Metal
PowerVR had PowerSGL
Matrox had MSI
Osiris
I permanently banned my 6 year old daughter from using the PS4 for mistakenly sending grief reports as it's too hard to watch or talk to her
Oh man, Rendition and PowerVR, boy do I remember backing the wrong horses! - Makes sense that I backed the Gravis Ultrasound too instead of the Soundblaster
(Although I remember the leap in 2D IQ from moving to a Matrox card around that time, was like cleaning vaseline off your screen and seeing it clearly for the fist time ever)
Glide was far from the only propitiatory 3D API, it was just the longest lasting / most successful until it's irrelevance due to DirectX catching up.
Sentenza
Member
Yeah, if anything having a lot of people in disbelief for this alleged 45% performance increase should be pointed as another reason to laugh at people claiming that "coding to metal" and other and console magic can get you 3X the bang-for-your-bucks or more.
That said, no one ever denied that you *would* have an increase in performance with low level access to the hardware, so I'm not sure who he was referring to.
It's also worth noting that this comes at a price in terms of effort necessary to get the most out of the hardware and expertise required for it, it's not a "free and easy" solution.
AMD opened the can of worms.
Maybe devs who're rolling their own rendering engines.
I dare say they may be in the minority these days, though?
And if middleware developers hate it - well, their job is to be the easy-to-use middleman on top of 'mess'
If, as a middleware-using developer, you know your engine will transparently (to you) give your game better performance for some people than they'd otherwise get, that would be a ++
Similar threads
- Locked