| Page 124 of 772 |
|
|
 Posted: Mon, 30th Sep 2013 17:27 Post subject: |
|
 |
It's just yet another API that only works on Brand X, just like PhysX, of course the competitors at Brand Y are free to use it too...... if they switch their architecture to that of Brand X, har har har. I was hoping we would be passed this by now, "features" that are brand locked.
|
|
| Back to top |
|
|
|
|
| Posted: Mon, 30th Sep 2013 18:13 Post subject: |
|
|
Not really the same thing, and it isn't really that simple either.
It's an API designed to allow working close to the "metal", removing overhead DirectX and OpenGL needs to be a general-purpose API.
It's not an addition on top of the existing API (like PhysX) - it's a replacement (for DirectX), if you choose to use it.
Simply put - you "create the graphics" with that API, like you would with DirectX 
It's a lot of work to build something like that from ground-up.
Because it's supposed to interface closely with the hardware, you can't simply make it work with other hardware - that would be making DirectX all over again, while greatly increasing the complexity.
I don't think it's about AMD "locking" it - Nvidia would have to write their own API either way. The only way I see it working for both, is if they collaborated from the start, and somehow designed their APIs to be reasonably easy to "port" from one to another. Some sort of unified dev tools maybe.
Mostly I would see this collaboration in solving some general problems and working on overall concepts when creating such an API - the implementation itself would still be up to the respective green/red teams.
The fact that is GCN only is not new, by the way
|
|
| Back to top |
|
|
Werelds
Special Little Man
Posts: 15098
Location: 0100111001001100
|
| Posted: Mon, 30th Sep 2013 18:40 Post subject: |
|
|
| sabin1981 wrote: | | It's just yet another API that only works on Brand X, just like PhysX, of course the competitors at Brand Y are free to use it too...... if they switch their architecture to that of Brand X, har har har. I was hoping we would be passed this by now, "features" that are brand locked. |
No, I'm not sure whether the article you linked misinterpreted it or you did.
First of all, it's not GCN only by design. The current implementation is, which makes sense because it's AMD developing it. They can't make GCN-exclusive either, because they'd be shooting themselves in the foot for their future architectures.
Second, if you look at the actual architecture (below) you see that there's a driver underneath the Mantle API. That is where the translation to the GPU's architecture happens. The actual API will be a bit more relevant to engine programmers. That is also the layer Intel or Nvidia would have to implement. Again, right now, there is only the GCN-specific driver. Compared to D3D or OGL, that removes the D3D/OGL driver, which in turn would translate to the OS driver, which in turn goes to the architecture-specific layer.
Third, they never said it was open like OpenCL or OpenGL. Those APIs are designed to be as generic as possible to support a very wide range of hardware. The purpose of Mantle is specifically to get as close to GPU shaders (which are fairly similar between all the architectures) as is reasonable, which is the exact opposite. It is open in the sense that the spec will be open to others if they're interested - Intel have already expressed their interest, for the record. Engine programmers won't be programming for GCN, they'll be programming for Mantle.
Fourth, Mantle is compatible with HLSL, D3D's shader language. That means porting is not *that* complicated for engine programmers.
Don't get me wrong, Mantle is all about AMD driving their own products. The fact that it accepts HLSL (Xbox One uses that too) and that it's obviously gonna have some things in there which work exceptionally well on GCN is a given. However, I've seen several articles out there misinterpreting both what was said at the conference as well as Raja's answers. It was never meant to be an open standard like OpenCL; it's an open spec, that's something very different.
And to draw a comparison to PhysX: that is a proprietary API, running off another proprietary API (CUDA) which is exclusive to Nvidia cards - that API in turn works directly on the metal, without a replaceable driver. With this one, it's a proprietary API with a public spec running off an extra layer in between that can be replaced (the Mantle driver). Had AMD really wanted to make this 100% exclusive, there wouldn't have been a driver in the middle.
The architecture:
 |
|
| Back to top |
|
|
|
|
| Posted: Mon, 30th Sep 2013 18:46 Post subject: |
|
|
Hey, I'm just repeating what the article said. Whether that's wrong or not I simply don't care, I'm just repeating - from where I'm sitting it sure as hell looks like another "exclusive" tech that is only applicable to Brand X. Just another way to force people into using their hardware otherwise risk missing out, I guess it's a "good" thing that my next GPU is planned to be a GCN then.
|
|
| Back to top |
|
|
Werelds
Special Little Man
Posts: 15098
Location: 0100111001001100
|
| Posted: Mon, 30th Sep 2013 18:50 Post subject: |
|
|
I realise the above may not make sense if you're not a programmer, but the big clue is that Mantle driver. PhysX and CUDA have no intermediary driver. The former requires the latter and the latter is exclusive to Nvidia. The Mantle driver isn't. Again, if AMD did not intend to at least have the option to let other architectures use it -regardless of whether it's their own or not-, there would not have been a driver there. That would've gone straight down instead and they would've just updated it for new architectures (or design new architectures around it) like Nvidia does with CUDA.
|
|
| Back to top |
|
|
|
|
|
| Back to top |
|
|
Werelds
Special Little Man
Posts: 15098
Location: 0100111001001100
|
| Posted: Mon, 30th Sep 2013 18:57 Post subject: |
|
|
I know 
I get your concerns and I do share them, but their conference and interviews do not confirm them. Only time will tell. Regardless, it's a good initiative and I can only hope they do allow others to create Mantle drivers - like I said, Intel has shown interest and that is a good thing. Intel does have greater interest in it than Nvidia though, because Intel's IGPs are in practically every laptop nowadays and something like Mantle could give those things a substantial boost.
|
|
| Back to top |
|
|
Sin317
Banned
Posts: 24322
Location: Geneva
|
|
| Back to top |
|
|
|
|
| Posted: Mon, 30th Sep 2013 19:07 Post subject: |
|
|
so as far as i understand it mantle api is better performance cause its DirectX and videodriver in 1 ? and the driver part is only programmed for GCN atm.
|
|
| Back to top |
|
|
|
|
|
| Back to top |
|
|
|
|
| Posted: Thu, 3rd Oct 2013 06:09 Post subject: |
|
|
Mantle interview with Raja
http://www.hardware.fr/focus/89/amd-mantle-interview-raja-koduri.html
| Quote: | | I think you may see more around developer summit. Mantle for us is a very developer driven effort, when they are ready they'll say it. Johan was ready he said it. It is not like we came up with Mantle then we kind of want to push it, it is not AMD's CUDA. |
Nice jab lol
|
|
| Back to top |
|
|
Werelds
Special Little Man
Posts: 15098
Location: 0100111001001100
|
| Posted: Thu, 3rd Oct 2013 13:01 Post subject: |
|
|
I sincerely hope that's the standard design, finally no black/red or black/green. This looks so much better tbh 
Also, review samples are being sent out now, meaning reviews will definitely be 2 weeks from now (15th), but we might get some real leaks before then instead of all these engineering sample nonsense results |
|
| Back to top |
|
|
Frant
King's Bounty
Posts: 24654
Location: Your Mom
|
| Posted: Thu, 3rd Oct 2013 13:38 Post subject: |
|
|
Mantle makes sense now that both PS4 and Xbox One (ffs MS, One??) use GCN-cores. Intel is definitely jumping on the train and nVidia will HAVE to jump on the train since the market is console driven and code on the consoles will be Mantle-driven.
I don't see any comparison to PhysX. I did the first second I heard about it but now it all makes sense. There's been many articles by various developers (carmack among others) about the problems with OpenGL and DirectX etc. being heavy API's that brings so much overhead that it makes it impossible to reach the actual capacity of the hardware no matter what. MS blocked developers from direct hardware access to get better compatibility many years ago. This is now an issue since efficiency-ratio from code to hardware rendering is hampered. Mantle makes the ratio a lot better.
Hopefully MS (DirectX) and OpenGL takes a hint and opens another lower layer to improve the efficiency without all that shitty overhead.
Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn!
"The sky was the color of a TV tuned to a dead station" - Neuromancer
|
|
| Back to top |
|
|
Werelds
Special Little Man
Posts: 15098
Location: 0100111001001100
|
| Posted: Thu, 3rd Oct 2013 14:28 Post subject: |
|
|
Well when DirectX was conceived, you also still had S3, 3DFX, ATI, Matrox in the same playing field; Nvidia was only just peeking around the corner then. MS targeted them all though, they wanted it to work for them all.
Now we've got AMD, Nvidia and Intel, where the latter is a follower rather than an innovator.
Also, 7950 now available here in NL for under 170 EUR. Prices are dropping like crazy 
|
|
| Back to top |
|
|
Sin317
Banned
Posts: 24322
Location: Geneva
|
|
| Back to top |
|
|
leroy15b
Banned
Posts: 293
Location: Dutch HigH Lands
|
| Posted: Thu, 3rd Oct 2013 15:39 Post subject: |
|
|
i have a Geforce GT 650M 1GB, hope i can run gta5 on medium..With 35 FPS! 
New laptop: i7-3630QM / 8GB / Nvidia 650M 1GB / 500GB
| NPiracy wrote: | | If I buy the game I actually on my way the little mine helpful and contribute to the company to develop. (GTA4/360) |
Last edited by leroy15b on Thu, 3rd Oct 2013 15:43; edited 1 time in total
|
|
| Back to top |
|
|
|
|
|
| Back to top |
|
|
Sin317
Banned
Posts: 24322
Location: Geneva
|
| Posted: Thu, 3rd Oct 2013 15:56 Post subject: |
|
|
|
|
|
| Back to top |
|
|
|
|
| Posted: Thu, 3rd Oct 2013 16:00 Post subject: |
|
|
|
|
|
| Back to top |
|
|
|
|
|
| Back to top |
|
|
Frant
King's Bounty
Posts: 24654
Location: Your Mom
|
|
| Back to top |
|
|
|
|
|
| Back to top |
|
|
Sin317
Banned
Posts: 24322
Location: Geneva
|
| Posted: Thu, 3rd Oct 2013 16:26 Post subject: |
|
|
ps4 and xbone are already outdated hardware wise.
NVIDIA made 170some million profit last year.
AMD lost 140ish million last year.
When NVIDIA say, it's not lucrative/worth it, i believe them.
|
|
| Back to top |
|
|
Frant
King's Bounty
Posts: 24654
Location: Your Mom
|
| Posted: Thu, 3rd Oct 2013 16:50 Post subject: |
|
|
| Sin317 wrote: | ps4 and xbone are already outdated hardware wise.
NVIDIA made 170some million profit last year.
AMD lost 140ish million last year.
When NVIDIA say, it's not lucrative/worth it, i believe them. |
NVIDIA doesn't have a massive haemorrhaging CPU division. The GPU-division @ AMD is very lucrative.
Now imagine the TWO top-consoles using AMD hardware. AMD can keep the margins low since the numbers will be ridiculous on sold APU's and profit. nVidia are REALLY sour over missing out this generation since they simply didn't have a viable solution to compete with.
And every console was basically outdated compared to the top-end PC at time of release, nothing new there. Consoles have an advantage, they're a completely homogeneous system and developers can get more out of a console than out of a PC since a PC has a thick layer of API's and interfaces before the first polygon has even been drawn.
AND... there are hundreds of millions of console-only gamers.
...never thought I'd "defend" consoles (well, I'm not actually, I'm just being realistic)...
Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn!
"The sky was the color of a TV tuned to a dead station" - Neuromancer
|
|
| Back to top |
|
|
Sin317
Banned
Posts: 24322
Location: Geneva
|
| Posted: Thu, 3rd Oct 2013 17:03 Post subject: |
|
|
NVIDIA CHOSE not to continue with consoles.
they CHOSE not to.
They didn't MISS OUT on anything.
They LET amd have that market.
I know it's hard to swallow, but that's how it is.
|
|
| Back to top |
|
|
|
|
|
| Back to top |
|
|
|
|
| Posted: Thu, 3rd Oct 2013 17:27 Post subject: |
|
|
| Quote: | Newegg.com has finally went live with the reference Radeon R9 290X from MSI that is priced at US $729.99. Although it is not "officially" listed but rather hidden in the HTML code, the price of US $729.99 suggest that the official price will most probably be set at US $699.99.
Since Newegg.com usually adds anywhere between US $10 and $30 to the suggested retail price it is anyone best guess that AMD intends to price the R9 290X at US $699.99. Newegg.com most likely wanted to keep the price hidden until last minute but HTML code shows that it will be set at US $729.99.
Bear in mind that AMD still can easily change the price at any time before the official launch and while many hoped for a $599.99 price tag, it looks like that AMD wants just a "tad" more money.
This price puts the R9 290X in a rather akward position as it might end up costing $50 more than the GTX 780. While the GTX Titan is still priced way above it at US $999.99 we still do not have a clear performance numbers to make the final judgement. Early details suggest that it will be faster than the GTX 780 but we honestly hoped that AMD might put a lot more pressure on Nvidia by pricing it below or at least at the same level as the GTX 780.
Of course, we still need to see some official reviews and prices, so we just have to wait a bit longer. |
http://forum.beyond3d.com/showpost.php?p=1791583&postcount=1017
What goes for the "reinforced" pcb, it seems to be using same kind of full cover VRM/Memory plate than GTX 580 (no bending at all).
|
|
| Back to top |
|
|
Frant
King's Bounty
Posts: 24654
Location: Your Mom
|
|
| Back to top |
|
|
Sin317
Banned
Posts: 24322
Location: Geneva
|
| Posted: Thu, 3rd Oct 2013 17:33 Post subject: |
|
|
why would it ?
Are you guys really so stuck up, that you can't believe that NVIDIA voluntarily decided NOT to pursue an investment they seem unfit ?
Is anything other then "AMD BEATS NVIDIA 11!1" impossible ?
NVIDIA's major market is in Graphic cards. For PC. That's where they always made most their money.
And i'm not saying consoles aren't profitable, heck, they sure are. The games that is. The hardware part is hardly profitable at all (why else would it take them 8-9 years for a new generation lol).
Console hardware is NOT profitable. Only Games are.
Unlike PC's, where the Hardware profits are probably on par with games lol.
|
|
| Back to top |
|
|
|
|
| Posted: Thu, 3rd Oct 2013 17:38 Post subject: |
|
|
There is a difference between simply deciding not to do it, and realising you can't do it at a profit. It's stupid to think that Nvidia would refuse if they could make money out of it.
AMD was in a better position with its hardware to actually pull it off. Nvidia apparently wasn't. And sure, yeah, AMD is not in a great place right now financially, and would really want to get that deal, but it doesn't mean Nvidia would want to simply let them. Not if they could help it.
|
|
| Back to top |
|
|
| Page 124 of 772 |
All times are GMT + 1 Hour |
