- Saturday, 7 November 2009
- INQ Mobile
- RSS
Why isn't New Year's Eve always on a Friday? - Adamson Rust
AMD won't let Nvidia's Cuda run on its GPUs
THE OLD COBBLERS that Nvidia is busily chipping away trying to adapt its Cuda technology to AMD's GPUs has emerged yet again, seemingly over some confusing comments made by the Green Goblin's chief scientist, Bill Dally.
In a roundtable discussion, Ben Hardwidge of Techradar asked Dally about Cuda, mentioning that it currently works only on Nvidia GPUs. "If you're a developer who wants to reach as wide an audience as possible, wouldn't it be better just to go with OpenCL?" probed Hardwidge, only to be told "In the future you'll be able to run C with CUDA extensions on a broader range of platforms." Dally went on to cryptically add "I'm familiar with some projects that are underway to enable CUDA on other platforms." He didn't elaborate further.
Surprised by this, the INQ decided to ask Nvidia outright whether it was indeed fiddling about with Cuda to allow developers to use it on both NV and AMD GPUs, making money off AMD's products. The answer, when it finally arrived after hours of waiting, was evasive.
Nvidia PR told us Dally had probably been referring to something else entirely, like a Linux-based tool designed to compile the CUDA programming model to a CPU architecture, or running C on anything from PCs, to handhelds, to servers and Playstations.
"So, nothing to do with AMD GPUs then?" *Cough, ehem, cough* "Er, we'll get back to you, but don't think so" - or something to that effect - came Nvidia's response. When that response half heartedly did come back to us, it stutteringly read "he [Dally] was giving a hypothetical....technology wise it could....both companies would have to do some work..." Aha. A hypothetical, eh? Hypothetically we could all be living on the moon by 2020 too.
Viewing the bigger picture of GPGPU, there are a couple of general purpose standards already available, including OpenCL and Microsoft's DirectX Compute, both of which are supported by AMD and Nvidia. But both firms have also decided to make their own proprietary flavours, with Nvidia far out ahead with Cuda, whilst AMD lags far behind with Stream.
What work AMD/ATI is doing is focused on Open CL. The implementation realities of Open CL, however, and whether it could support heterogeneous ATI/NV mixed environments are still very much unclear.
Cuda is generally believed to be a fair bit better than Stream - although, obviously AMD begs to differ - with a multitude of developers having already put it to the test on hundreds of applications, which are already available. AMD, on the other hand, ‘boasts' a paltry five ISVs, two of which are also working with Cuda, and says a development driver for OCL will be out "Very soon". Nvidia's has already been released.
Still, for Cuda to be able to work on AMD GPUs, Nvidia would absolutely need AMD's support. Without it, Nvidia wouldn't be able to get low-level programming access to the GPU to develop the API. Even Nvidia admits that AMD would probably never allow this to happen. As for AMD, the company's point man on Stream seemed amazed we'd even asked.
AMD's Gary Silcott told the INQ "they [Nvidia] would intentionally damage performance to make Nvidia GPUs run the same app better." Then, perhaps thinking better of accusing Nvidia of hypothetical, yet outright, sabotage, Silcott added "Even if it wasn't intentional, it would not be optimized for our instruction set architecture like our own SDK."
That's okay though, since Nvidia has no intention of adapting its GPUs for AMDs technology either. "No, I don't see us supporting Steam..." said Nvidia's Derek Perez acidly when we asked him for his response. µ
Share this:
ShareBusiness Intelligence Software
Customer Relationship Management (CRM)
Enterprise Accounting Software
Enterprise Resource Planning (ERP)
Enterprise Systems Management
Tired of the video card fight's. When i need a video card i just get the best bang for the buck. Last video card i got was $69.99 ATI HD 4650 1 Gig mem it was cheep and does what i want it to do. No more 200,300,400,500$ video cards for me. And i am for open standards.
DAAMIT can sell more card if they let Nvidiot do what they wana do. I mean why not? Nvidia already granted Intel the use of SLI so what the big deal w/ a company that is not financially stable? I mean Hector left and the new guy still narrow minded? What a bunch of monkeys! Shame on AMD and I still hope that they will overcome stupid ways of thinking or the ugly shadow of bankruptcy will be upon them!
DAMMIT says: Nvidiot leave my cards alone!!!
Nvidians say: CUDA works better in you DAMMIT! lol
I know I know, is all about money.
DAAMIT is living in fear, I bet they don't want CUDA because it shall focus more attention on the differences between themselves an nVidiot!
A direct comparison of how CUDA runs on cards from both vendors with various applications is the last thing DAAMIT wants. Sticking with gaming statistics and relying on them for performance testing is prefered. As you can at least influence the programmers in the development stage to take advantage of 'your products special features'. There for creating a variable world, where marketing scum loves to thrive and spread there filth. Trying to put spin on Photoshop etc other than yet another game is a lot harder... Bring on CUDA!
I'm quite bored with the comment section when it comes to graphics. Everyone overuses exclamation marks !!!!!! in defense of their preferred graphics company. If you want to make a comment, please, calm yourself, and make intelligent arguments in a relaxed and wise manner, instead of screaming your point of view as though you think if you're loud enough, you may win the argument.
I suggest Rich Evans, that instead of complaining!!!! about overuse of said!!! then you go to a library and grab a book!!!! No one is screaming over anyone!!!! you sound like a dull man with nothing more to whinge about than the comment section on here!!!! I quite like the use of!!! there should be more of it!!!!!!!!!!
I can see these vendor specific APIs dying before they're even weaned, so the point is moot anyway. OpenGL, OpenAL (not Khronos' but put together in the same spirit as the other two) with OpenCL to complete the set. All three have open and published specifications.
In short, the two (three if Intel's Larrabee appears on time) warring parties have hardly anything to gain from this little battle as the Khronos Group has the advantage of being trusted, open, fairly free with documentation and platform/OS neutral. Or, if you prefer the sensationalist headlines, had their target market by the knackers as soon as they released the OpenCL draft.
The one and only company that could save CUDA or Stream is Microsoft, by adopting either. I very much doubt this is going to happen, simply because MS would be alienating a good chunk of its users no matter which of the two it chose. MS is more likely to add a fourth competing API (probably buried in its DirectX package) and force manufacturers to support it as a condition of WHQL certification than adopt someone else's.
That is AMD's position, pure and simple. The idea that Nvidia would pull some dirty tricks is a credible one, but it's all about using ATI's market share as a wedge to push developers away from CUDA.
I for one look forward to OpenCL.
"AMD's Gary Silcott told the INQ "they [Nvidia] would intentionally damage performance to make Nvidia GPUs run the same app better." Then, perhaps thinking better of accusing Nvidia of hypothetical, yet outright, sabotage, Silcott added "Even if it wasn't intentional, it would not be optimized for our instruction set architecture like our own SDK."
Err..... running Apps Better is what the Main Great Game is All about, Gary, and one would reasonably expect that to be a Funding Priority for Global Foundry Investors/Sleeping Partners. They wouldn't/shouldn't be too pleased with such a Negative Sub-Prime Approach to a Lead Competitor in Graphic Computation for Head Quarters.
Seems like AMD are missing some Vital Intellectual Property and may therefore be Abusing Arabian Hospitality and Ingenuity?
Given that AMD have open sourced their GPU specs so that they can pass the burden of driver development onto the Unix community, is this really true?
I recall reading that nvidia will continue development on CUDA and also support the OpenCL standard. Nvidia first to release driver that support opencl. OpenCL is the just vanilla version while CUDA is the vanilla plus the extra chocolate topping on it. imo, I think nvidia choose to support both cause open standards always have a longer cycle between each new version. And nvidia likes to release new revision every few months. So by keeping CUDA, they can update CUDA a lot quicker than making new revision to OpenCL. Do I like it that CUDA is proprietary? No but then all companies have propreitary tech. Intel with their x86 license, AMD going all inhouse: CPU and GPU in fusion, APPLE.. those guys is the king of propreitary. So people bashing nvidia for cuda, yet buys Intel CPU or planning on Fusion but own a ipod, etc... is just hypocrites.
I am not surprised AMD is not willing to give Nivida low level access to its GPUs. Sabotage seems much less likely than poorly optimized. If Nvidia can optimize AMD GPUs better then AMD seriously need to fire their team and/or higher Nvidia's.
OpenCL looks to be much superior than CUDA with ability to load balance between multiple computing resources although I have not yet tried OpenCL.
I see this as an attempt to keep CUDA relevant with the advent of OpenCL.
I think they should both see the forest before the trees. The fact remains that they both share a common enemy that could be defeated if GPGPU becomes mainstream. Let's face it, a GPU smokes a CPU in raw numbers. If I had a choice between a netbook with via processor coupled with either a nvidia or ati GPU, or an intel with integrated graphics, I would be on the fence. But if Nvidia and AMD/ATI standardized either CUDA and/or OCL the choice would be simple. I really like both Nvidia and AMD/ATI, just went from an 8800gt sc to an HD4870, and if the price is right (I paid 129.00 for the HD4870 opened box-couldn't pass it up)I wouldn't hesitate to get a GTX280.
They really need to make the GPU the center of computing before intel makes a decent GPU.
AMD is only surviving. They can't grow big cause Intel will always be the big monster that will knock it down. Intel will give AMD a small piece of the market but only enough to keep the government dogs away. If AMD wants to break out of this cycle, they should work with nvidia instead of sitting on the side and waiting for them to fight it out. If nvidia loses and becomes smaller and intel gain a good foot hold into the gpu market, then it is like history repeating itself again (with AMD next). Like back in the days when AMD let Intel grew it revenue quicker than AMD could. Eventually Intel became the big monster it is today.