Yesterday, AMD revealed Mantle, an API with which developers can access graphics hardware more directly, and one that Microsoft and Sony could end up being "downright hostile" toward, according to a tweet from id Software co-founder and Oculus VR CTO John Carmack.
The idea behind Mantle is to give developers low-level access to the native language of AMD's Graphics Core Next architecture, thereby easing the process of optimization for the company's Graphics Processing Units. The theoretical advantage is that, as opposed to high-level APIs like OpenGL and Direct3D that allow broad optimization over a wide swath of GPU hardware, developers can customize their products specifically for AMD hardware.
Coupled with Valve's announcement yesterday that the company will be "working with multiple partners" to create hardware running the Linux-based SteamOS , AMD's announcement which may offer developers the kind of hardware optimization options that exist when developing for hardware-locked consoles may create tension between Microsoft, Sony and AMD because of next-gen hardware.
I'm sticking to what I said in the other thread(s) - Mantle will have about 5-10 games per year that support / run native. Beyond that, it will either add a few extra little effects (or increase poly counts) and eat up the performance, or it will add no additional effects and have a small/modest performance bump.
People dreaming of 30-50% are doing just that - dreaming.
I'm sticking to what I said in the other thread(s) - Mantle will have about 5-10 games per year that support / run native. Beyond that, it will either add a few extra little effects (or increase poly counts) and eat up the performance, or it will add no additional effects and have a small/modest performance bump.
People dreaming of 30-50% are doing just that - dreaming.
AMD only needs to do 3 things to get mantle to take off. Get it in Frostbite 3.0 (mission accomplished), get it in UE4, and get it in Cryengine. They do not need to target individual games, or studios. IF they get it in the big 3 engines they will cover the majority of multi-platform games being released next gen. The days of dev's making their own engines are done. Game development is too expensive. They are going to the big engine devs because of the automation they build into their engines which saves them money and time which they can use to polish their games.
If mantle is put into those engines and works automatically (which is the end goal here I imagine) there is literally no extra work that needs to be done to use it. Your games are automatically optimized for both next gen consoles, and as a side effect also GCN equipped PC's.
If mantle is put into those engines and works automatically (which is the end goal here I imagine) there is literally no extra work that needs to be done to use it. Your games are automatically optimized for both next gen consoles, and as a side effect also GCN equipped PC's.
AMD only needs to do 3 things to get mantle to take off. Get it in Frostbite 3.0 (mission accomplished), get it in UE4, and get it in Cryengine. They do not need to target individual games, or studios. IF they get it in the big 3 engines they will cover the majority of multi-platform games being released next gen. The days of dev's making their own engines are done. Game development is too expensive. They are going to the big engine devs because of the automation they build into their engines which saves them money and time which they can use to polish their games.
If mantle is put into those engines and works automatically (which is the end goal here I imagine) there is literally no extra work that needs to be done to use it. Your games are automatically optimized for both next gen consoles, and as a side effect also GCN equipped PC's.
I don't know if Mantle will take off or not, but I think it has some things in place that give it more of a fighting chance than some would like to admit.
1. Frostbite will more than likely be limited entirely to EA games.
2. UE4 is squarely in Nvidia's camp. It has physx integrated into the engine.
http://www.polygon.com/2013/6/6/440...ted-partners-program-oculus-rift-nvidia-physx
3. Whether or not Crytek will support it is another thing. I just don't know, but I can see Nvidia shelling out money to developers to prevent mass migration.
4. If Valve ever comes to market with a next generation Source engine (and Gabe has been quoted as saying it is in the works) it is more likely to see third party games on it than Frostbite. That said, I doubt Valve will support Mantle when they just built SteamOS and are seemingly working with, but not limiting themselves to, Nvidia (hardware specs on Valve's personal Steam Machine prototype are to be announced next week).
I just shot down 3/4 of what you think Mantle requires to take off.
Agreed but in the end marketing, manpower, and resources are going to be required to get Mantle proliferation. Do you think AMD is in the position to adequately provide all of these things?
I believe we are forgetting the sleeping giant here, Intel.They already have a world class compiler, they just need to persuade the devs to use some newer instructions(which I believe they can do better than anybody else). Now if Intel is at all interested in pc gaming is another mater entirely.I was pleasantly surprised to see Total War Rome 2 as a Intel sponsored game.
And all of those developers have GCN as their main target starting in November. They all want to make money, so they will be targeting multiplatform releases. They all want their game engines to perform the best they possibly can.
1. EA has 15+ games coming on Frostbite for next gen consoles.
2. UE4 is squarely in nvidia's camp, but what happens when no one licenses it because being squarely in Nvidia's camp as you say they optimize for Nvidia hardware only? Oh right, that will never freaking happen because they are forced into targeting GCN gpu's by default due to the consoles. When 100's of studios are screaming at Epic for the ability to easily move their code back and forth across PS4, Xbone, and PC they will have to capitulate. Can nvidia afford to buy their loyalty in comparison to the amount of money they receive for engine licensing?
3. Crytek is published by EA, see #1.
4. Valve has stated that they are working with multiple partners on linux. AMD says mantle will work in linux. Mantle is announced the same week as Steam OS. Valve wants to end Windows gaming and DX reliance, and lucky for them AMD has done just that, and made it compatible for Consoles, and PC's. Did Valve release any games on console last gen, I can't remember?
I just shot down 100% of what you said.
Yes it will be a lot of marketing,
Agreed but in the end marketing, manpower, and resources are going to be required to get Mantle proliferation. Do you think AMD is in the position to adequately provide all of these things?
There has been nothing stated that Mantle will cripple Intel CPU's. To the contrary it should make them perform better in games.
I'm sticking to what I said in the other thread(s) - Mantle will have about 5-10 games per year that support / run native. Beyond that, it will either add a few extra little effects (or increase poly counts) and eat up the performance, or it will add no additional effects and have a small/modest performance bump.
People dreaming of 30-50% are doing just that - dreaming.
He did a test in CryEngine to see how draw calls effected his FPS.KKRT00 @ http://www.neogaf.com/
There are instanced where it could help tremendously, but game would have to be designed specifically with that in mind.
Btw i've just made draw call performance test in CryEngine SDK, quite surprising results
Video - http://*********/GrSpm2AZWVU (draw calls are listed as DP 3rd row from the top)
Results, not from video, but a little more precise testing:
300 draw calls - 105 fps
2100 draw calls - 104 fps
3000 draw calls - 103 fps
4000 draw calls - 101 fps
5000 draw calls - 91 fps
6000 draw calls - 83 fps
7000 draw calls - 75 fps
9000 draw calls - 65 fps
13000 draw calls - 49 fps
17000 draw calls - 41 fps
20000 draw calls - 37 fps
on stock i5 2500k and GTX 560. Dunno if my GPU has anything to do with it, but recording with FRAPS havent affected fps in any meaningful way.
My optimistic claims are mainly because this is what I see that AMD absolutely HAS to do to survive. Their very existence hinges on this if you ask me.
They will either make it happen, or cease to exist.
I honestly have no idea... on the surface I would probably say 'no' to all three. But it seems AMD has really made gaming a priority as of late, I don't think any of us would have thought a year ago Gaming Evolved would have scored the big name games it has, AMD would be providing the graphics for all three next gen consoles (and CPU's for two of the three), or would be rolling out their own API. It makes me wonder if Sony and MS both were happy to go with the less than powerhouse Jaguar cores because AMD already talked Mantle to them (speculation on my part... less than speculation really, more of a 'what if').
I feel Physx never went anywhere not so much because it is only officially supported by Nvidia GPU's, but because it never had a killer app, it seemed to always be there as an afterthought or checkbox feature. AMD already has BF4 on board. If Mantle really does what it is supposed to do, and BF4 really looks leaps and bounds better than the competition then where does that leave the other game developers? They'll have to get on board, build something that runs as well and looks nearly as good with conventional tools, or compete on a price and/or 'our game will run on your hardware!' position. I don't think Crytek is big on the tag line "Our shooters look second best!"
But there are a lot of ifs along the way. I think at the very least I would like to see a game running on Mantle before I say how I think it will do in the market place. If graphics and performance really are jaw-droppingly good, then I'll likely be much more bullish on Mantle then as it sits right now today, little more than a buzzword.
I googled "AMD + Mantle" and found some other forums talking about it.
Found this post intresting:
He did a test in CryEngine to see how draw calls effected his FPS.
Now most games wont use anywhere near 20,000 draw calls.
Lets say the most extreme games today use like 7,000 draw calls.
You can see his i5 2500k, gets about 75 FPS at 7,000 draw calls.
Now if there was suddenly a 9x speed up in draw call performance?
well, his i5 2500k would handle those same 7,000 draw calls like they where performing
at a rate of 7,000/9 (the x9 speed up) = ~778.
You can see at around 2100 draw calls (more than 778) he gets about 104 fps.
Thus in this case a 9x speed up of draw calls would result in him going from 75 fps -----> 104 fps.
Thats about ~40% increase in performance.
This might be optimistic but its realistic that performance incrases from this could end up in certain situations around that area.
What??? Intel is already demolishing AMD and they definitely don't want them to go any faster.It would make AMD's cpu a laughing stock.