• Happy holidays, folks! Thanks to each and every one of you for being part of the Tom's Hardware community!

Carmack: Microsoft, Sony May Be Hostile to AMD's Mantle

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
Carmack talks big and fells short most of the time, nothing new here.
1. As already stated in this thread. MS, Sony, Nintendo are deffinetely know about Mantle, they bought AMD products and enhanced them but AMD isn't stupid to lose such partners and wouldn't be silent about new API, thats obvious. Platform holders sell games, they know PC hardware is stronger, they're not stupid or blind, they count money ver well.
2. Before it was like this PS3 had custimized OpenGL + their own teams code to hardware, XBox360 had customized DX + their close to hardware approach, same with Wii. PC used Direct3D. Almost all of the teams either have their own engine or license the only engine thats used in game industry that will(already) likely have support for Mantle.
So it was 1 PS3 version, 1 Xbox version, 1 Wii version, 2 PC versions = 5 different versions!
Now it will be 1 version(PS4, X1, WiiU, PC[AMD]), 1 PC[Nvidia] = 1 Mantle + 1 Direct3D.= 2 different versions!
Developers will use it. First, code with Mantle support, then if interested add some better support for consoles while PC[AMD] is ready already. Less work for devs = more quality for us OR faster releases.
 
I Like this. Maybe 99% of the games wont use hardware level language, but monsters like Carmack and similar can create pure awsomeness with it.

Its like giving you the chance to fly. You dont have to, but its awsome to have the option.
 
Since NV owns 65% of the discrete market, I don't see anyone writing for this unless paid directly by AMD, who has no money to push an NON open standard. I'm not even sure NV could pull this off. It took what, 7 years and a lot of funding to get CUDA to become standard and in every pro app out. AMD can't fund that kind of time and effort. This is a dead tech, but nice try...LOL. I'd rather have money spent on better drivers so I'm maybe not stuck buying NV by default once 20nm hits just from fear of crap drivers for a few years (like now). GET BACK TO YOUR CORE AMD! Start making GREAT cpus and drivers again. Of course to do that, you have to fire the guy who keeps thinking you can PRICE NV to death via cheap cards, and also the guy who keeps telling you to give away all your profits via FREE GAMES with every card you sell. Fire those two and maybe you'll start making some cash instead of losses and selling every piece of business you have.
 
Apart from the fact that consoles do not perform as well as PC any more (not for a long time), even so, my perception has been that you get people who prefer consoles and you get people who prefer PCs. I really can't see mantle making a massive impact in that basic division of the market. A decent PC is still very expensive compared to a console and consoles still require very little user configuration if any - convenience. So the console people are going nowhere.

Mantle will just basically allow AMD to severely out compete NVidia on performance on titles that support this API until NVidia starts supporting it. AMD, unlike anything NVidia "the Apple of computer graphics" would do, made it opensource.
 

Mantle may reduce API call overhead by ~90% but that does not make the underlying hardware any faster and that overhead accounts for maybe 20-25% of system time. This allows developers to either free up ~20% of the CPU by reducing overhead or make 9X as many API calls per second to make more finely grained calls that might enable the GPU to work more efficiently.

As for Nvidia supporting that API, that seems unlikely since Mantle is suppose to give low-level access to GCN resources - this means non-portable almost raw access to architecture-specific features. Unless Nvidia starts making GCN-based GPUs, porting Mantle to Nvidia or even AMD's own non-GCN GPUs is likely impractical if not impossible.
 
I think mantle would be also especially useful for supercomputers or other uses where you are working with one gpu architecture anyway.
 
In consoles the Mantle is very usefull. It would reduce the time to port the game from one console to another considerably, because the same graphic API could be used in both consoles. No problems in there!
It is just a tool to use AMD hardware in really optimised way. In PC it could cause havok, but that remains to be seing. Most games will stay using DX and open GL, but some crosplatform games are in the spot. Interesting to see how it will be used.
 
I see Mantle as helping AMD's APU business more than mainstream or high end Discrete GPU's. Though I'm betting anyone who has a discrete GPU based on GCN will likely benefit by default as well. I'm running a 2GB HD7850 for example. I believe it will also have to do with how well the low level mantle code, and mantle portion of the drivers interface with DX and OpenGL. Since if they interface well, it may work out even better for us as PC gamers.

Now on the 3dfx and glide part.. Not many people realize, that if 3dfx hadn't shot themselves in the foot after the Voodoo Banshee, and decided to turn to in house board manufacturing, they'd likely still be around. When they cut off the ability for outside OEMs like diamond, STB(use to have an STB Black Magic Voodoo2), MSI, Creative Labs, etc. And went In house, competition ceased, and price on 3dfx based graphics cards sky rocketed. If anyone remembers, the Voodoo3, and Voodoo5 cards were both made in house by 3dfx, and both were super expensive. Especially when they were compared to competing Nvidia and matrox based cards. Along with Voodoo not having support for 32bit color at the time, thats what actually killed 3dfx and glide, DirectX adoption had nothing to do with it. Simply put, people weren't going to spend more on a card, that didn't always perform that much better than competing cards.
 


https://linustechtips.com/main/topic/60597-amds-mantle-api-is-not-open-source-compatible-only-with-amd-gpus/
Mantle IS NOT OPEN SOURCE - won't work on NV cards. So says AMD's Robert Hallock. It only works on GCN!

These two companies are the same. Neither does anyone favors, and both are out to make money (though one seems to know how to do this part better than the other). Both are evil :)

https://twitter.com/Thracks/status/383872285351739393
Source of Robert Hallock's post. Direct response! NO NVIDIA. So just as evil as that TERRIBLE NVidia right?...ROFL. NV haters...They crack me up.
 

Open source (availability of source code) and open standards (portability between software and hardware vendors) are two different things.

You can have open source stuff that remains fundamentally proprietary when the functionality it provides is tied to a binary blob or hardware interface that has no practical common standard representation... such as a GPU architecture with its data structures and instruction set.

Pretty difficult to provide hardware abstraction layers to an API whose sole reason for existing is to expose raw proprietary hardware to programmers. The only way to "port" a low-level API to other GPU/IGP designers would be for Nvidia, Intel and friends to license the GCN architecture from AMD and start making GCN-based GPUs/IGPs.

With low-level APIs, the hardware defines the API or the API defines the hardware. Either way, you end up bound to a specific vision of things and in Mantle's case, that means AMD's.
 


What's your point? The guy asking the AMD guy (in the links I provided) asked if it was OPEN SOURCE, to which Robert Hallock (AMD guy) said NO. The statement I responded to said it was OPEN SOURCE. Hence my response.
"Tech Fan@ic . ‏
@GnrlKhalid 27 Sep

@Thracks is MANTLE open source ?
Robert Hallock ‏@Thracks

@GnrlKhalid No. It is an API for the industry-standard GCN Architecture and its specific ISA, done at the request of game developers."

But I can take it a step further if you'd like and say it's neither open source (not aware of them handing out the code for others to chew on), NOR is it an OPEN STANDARD (it really can't be, since it's tied to AMD hardware, and only GCN at that).

I realize they are two different things...But in this case care not no matter how the other guy takes my response, it's still correct (the AMD guy got what was meant by the question). :) The only thing needed here is a clear answer stating it's closed in every way possible :) NV with 65% market share of discrete cards, and 90% of the pro market won't be signing up to make GCN cards even if AMD allowed it.

Microsoft/Sony won't help this either as it only makes directx weaker (and gets people possibly off new consoles eventually) and strengthens other platforms for games (steamos/linux/android etc). This is dead as I stated before. It is rumored it cost AMD 8mil to get Battlefield 4 optimized. How many times can they afford that with a video card division that made ZERO last Quarter? I really hope they didn't spend much on Mantle or the Audio crap. They are both losses. More coding for a specific platform=more cost to devs, so if you're not paying them directly, and MS/Sony is blocking you, and 2/3's of the market (NV owners) and your own NON GCN cards can't play it anyway, this IS DEAD. I know you're not making an argument for the Mantle API above, just making the point in case anyone is still thinking Mantle will change the world...LOL.
 

Neither link you supplied actually specifically state whether the API itself will be open-source or not. All they do say is the same thing I have been saying: bound to a specific hardware architecture and therefore not portable to anything else.

Too many people are confusing open source with open standard but the two terms are neither mutually inclusive or exclusive; all four possibilities are common in everyday life.

It is certainly possible to have an open-source API like Mantle to a closed-standard like GCN. The API itself is open but useless for non-GCN hardware.
 


So if it's bound to YOUR specific hardware, and nobody in their right mind would help your hardware by making it themselves in this case (Intel or nvidia, the only other 2 GPU producers for windows really), it's closed no matter how you put it right? We can argue semantics all day, but at the end of the day IT'S CLOSED right?

I'll give you that it's POSSIBLE Nvidia or Intel might make an AMD based architecture card. But at the same time I'll say I'll eat my hat, my car and my house & my neighbors house if that ever happens. Which should be about right after hell freezes over, pigs fly etc.etc....ROFL. :) So I say there is only one possibility. AMD either figures out how to pay every game maker to do this (apparently around 8mil...) or realize it was wasted R&D because nobody wants to write 2 sets of code for free. They'll do the one that covers everyone and even that is a risk if the game sells like crap, but they won't do the one for YOUR cards only without payment.

Even if some cards are not fast enough to play today, next year (year after?) everyone with slow ones can replace their cards (if they want to play your game bad enough they do it) with the next die shrink and play them most likely unoptimized for a specific makers gpu. So why would I even waste the time/effort/cost? I might not say that if it was AMD who had 65% market share, but that isn't the case so you're likely not going to do free extra work for a the smallest audience and even then not AMD's whole audience as it's only GCN cards not older ones. Everyone can play Crysis 1 now right, and they didn't have to do anything but wait for it to happen. :) No extra cost or risk, they just waited it out.
 

Console GPUs and CPUs won't be upgraded for another 5-7 years. As I have been saying from my very first post on Mantle, that's where I think Mantle's primary purpose lies and PC gamers who have GCN-based GPUs simply get that as a freebie - it does not cost AMD significant extra R&D cost to port Mantle from consoles to PC since most of the work is already done for consoles.
 
Dear Carmack,

did you remember how crappy your RAGE port was from console to PC? These were your exact words,

"The driver issues at launch have been a real cluster !@#$"

See you soon @ Doom 4 launch


Your fan,

Annoynimous
 


Just looking at the 2800 devs that showed up at GDC 2013 (2500 US, 300 Euro), games won't be made this time on CONSOLE first. Only 11% are planning anything for them, note the PLANNING, not making now. Meanwhile 40% planning PC, 60% planning mobile. This is creating a chicken egg thing. Consoles have to sell well enough in the first year to get people BACK to making console first. Last gen, devs thought they'd make their money back eventually on any game over the 7-8yrs. They don't believe that this time and have serious doubts as GDC survey shows. For the first time games will be made elsewhere first and MAYBE make it to consoles once enough sell to be worth porting to them. I don't think consoles even matter this round, they are dead and the writing on the wall says Directx is right behind them (which is basically valve's whole point and goal, Google's also). Valve/NV/Android/Google (apple/amazon too?) all have perfect timing to end consoles once and for all.

http://www.gdconf.com/news/gdc_state_of_the_industry_rese.html
Interest in consoles is dropping while other areas are exploding. A large portion made their last game on Mobile first, and even more planning that on the next game. The opposite is true for consoles.
http://www.insidesocialgames.com/category/gdc-2012/

You can see they have higher INTEREST, but that's just jargon for wait and see...NOT making anything. We will see if consoles can get them from 11% up to that 27/29% interested number. But for now, what is being MADE currently is mobile/PC. Note PC/TV consoles (steambox mentioned) are at 45% interest, probably because a good 1/4 of us already have steamboxes in our current PC...LOL.

I think Valve just announced specs so people at home can get a rough estimate of it they already have one in house or not, so they won't buy a console. There are millions of people with 660's and up (7950+ too) so it's already a larger market than consoles will be on next gen for at least 2yrs probably. Valve just needs that OS installed and we have millions in the market. I'm sure this is what the 45% interested in these boxes is looking at (instant numbers with just a steamos install). Everyone knows what to buy at xmas now, assuming a 660/i3 combo is ok already, just aim there or above and install steam to avoid a console. Nice move valve. :) Sure they'll have boxes of all shapes, sizes, power etc launched for people with no PC knowledge for off the shelfers. But the main thing I think was to let us know if we already own one or what's needed for an xmas gift to make one :) Most of use just needed a gpu recommendation as many are already i3 or better sandy/ivy/haswell.

You're also assuming (wrongly IMHO, Carmack's opinion too), that sony/MS will welcome Mantle, but they won't. It's a threat to them, not a bonus. Games optimized for mantle can be ported more easily with mantle crap to elsewhere (linux, steamos, android etc) making the need for consoles & directx less over time. You want your console game not easily ported to anything but PC's WITH Windows or your platform shrinks. Imagine 5yrs from now 90% of games being linux & windows because of mantle support (instead of the 10% available now). Do I need windows then? Consoles? I can switch and still play on my TV. They will not help Valve kill them, they will fight anything the helps Valve, Android etc diminish their own platforms.

Mantle will be BLOCKED for PS4/Xbox1. AMD might have pulled this off 10yrs ago getting it into last gen. Now though, there are a dozen consoles out now or coming shortly, that would like easy ports from consoles or windows. They will not help these survive. Helping Mantle helps OTHER consoles succeed (easy for devs to jump to others then right? - don't want that if you're sony or microsoft). They don't see PC's as competition (quite yet, though streaming to any crap device to TV might make them think they are soon), so it's SEMI-OK to make PC/console ports easier for now. But I think that's changing as NV already streams, Valve helping, and AMD is working on it to.

When they started the idea of making games easier to port from console to pc in this gen it was on the table ~5yrs ago, not knowing everyone would be trying to stream from PC to every smart device on the planet and then to TV. I'm guessing they are now sitting around that table thinking "JEEZ WE'RE STUPID, WE JUST KILLED OURSELVES". Streaming low latency to TV will be conquered shortly (NV almost done proving this locally with shield from PC), and I wouldn't be surprised if they get it doable for most games from the web too (all but the most sensitive games). So now [with game devs fearing consoles may never sell enough] instead of games being made Console first, they come PC or something else first, then consoles never matter to begin with this gen. OOPS. :)
 

Well, no, actually.

Steambox uses Nvidia chips so no Mantle support there. Most other me-too consoles use ARM-based SoCs so no GCN nor Mantle on those either.

If you remove Mantle, what do console developers fall back on? DirectX for Microsoft and OpenGL for Sony which are much more more generic and portable than Mantle as long as programmers do not bypass APIs to wring more performance out of consoles.
 

One thing to keep in mind: support is always only one firmware update away should Microsoft change their mind... and the XB1 with its weaker GPU might benefit more from faster and more finely grained GPU calls than the PS4.
 
well this article are talking about Carmack statement that MS and Sony could be hostile towards AMD Mantle. so isn't this somekind of confirmation that MS doesn't want Mantle related stuff when develop for their new console 😛
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom