AMD Launches Boltzmann Initiative - HSA porting CUDA-based applications

Published by

Click here to post a comment for AMD Launches Boltzmann Initiative - HSA porting CUDA-based applications on our message forum
#5190181
https://forums.guru3d.com/data/avatars/m/174/174929.jpg
Bitcoin mining?
#5190201
https://forums.guru3d.com/data/avatars/m/229/229509.jpg
Heh, big middle finger to nVidia...
#5190205
https://forums.guru3d.com/data/avatars/m/258/258664.jpg
I have to say a very fine move on AMD's side if truely working without any performance losses. Take away an enemies strength if you can't beat them by yourself 😀
#5190210
https://forums.guru3d.com/data/avatars/m/169/169351.jpg
Maybe we'll get PhysX on AMD GPUs in the future then?
#5190213
https://forums.guru3d.com/data/avatars/m/174/174929.jpg
Heh, big middle finger to nVidia...
Aint that the truth...
#5190228
https://forums.guru3d.com/data/avatars/m/80/80129.jpg
Nvidia standardized/licensed the language to allow anyone to write a compiler for it, specifically for this reason. So I don't see how this is a middle finger to Nvidia, it's literally what they intended.
#5190234
data/avatar/default/avatar38.webp
Nvidia standardized/licensed the language to allow anyone to write a compiler for it, specifically for this reason. So I don't see how this is a middle finger to Nvidia, it's literally what they intended.
Its a bit more than that.. a more in depth article: http://www.anandtech.com/show/9792/amd-sc15-boltzmann-initiative-announced-c-and-cuda-compilers-for-amd-gpus As for the conclusion...
there’s a question of just what NVIDIA’s reaction will be. CUDA is NVIDIA’s, through and through, and it does make one wonder whether NVIDIA would try to sue AMD for implementing the CUDA API without NVIDIA’s permission, particularly in light of the latest developments in the Oracle vs. Google case on the Java API. AMD for their part has had their legal team look at the issue extensively and doesn’t believe they’re at risk – pointing in part to Google’s own efforts to bring CUDA support to LLVM with GPUCC – though I suspect AMD’s efforts are a bit more inflammatory given the direct competition. Ultimately it’s a matter that will be handled by AMD and NVIDIA only if it comes to it, but it’s something that does need to be pointed out.
#5190250
https://forums.guru3d.com/data/avatars/m/243/243702.jpg
Maybe we'll get PhysX on AMD GPUs in the future then?
No, you need source to compile library. It is quite possible that this compiler will be free as most of AMD development tools. But one requires source PhysX CUDA code, that's in nVidia's hands. That 2 years old DirectCompute promise... I do not think nV will make PhysX available to run on AMD hardware even if it was effortless to them.
#5190266
https://forums.guru3d.com/data/avatars/m/175/175902.jpg
Bitcoin mining?
still use that crap?
Heh, big middle finger to nVidia...
right now... no, as the green are much more efficient and AMD is still anecdotic in pro station... maybe in the future...
I have to say a very fine move on AMD's side if truely working without any performance losses. Take away an enemies strength if you can't beat them by yourself 😀
...wich is not so evident right now.
#5190293
data/avatar/default/avatar40.webp
Bitcoin mining?
Bitcoin miners already exists in OpenCL variants, and were typically always better on AMD hardware than the Cuda ones on Nvidia hardware.
#5190297
data/avatar/default/avatar34.webp
still use that crap? right now... no, as the green are much more efficient and AMD is still anecdotic in pro station... maybe in the future... ...wich is not so evident right now.
Historically AMD GPGPU performance was well above that of Nvidia's. This didn't see a significant change until Maxwell.
#5190380
https://forums.guru3d.com/data/avatars/m/248/248627.jpg
I just hope this pays out for them they need market share bad
#5190409
https://forums.guru3d.com/data/avatars/m/90/90034.jpg
I think this is a really good move, AMD GPUs have had the advantage in GPGPU for a good while now but it's not really been pushed forward in any way. Not being CUDA compatible is a pretty big hindrance when it comes to appealing to the professional market, this should change things up significantly!
#5190438
https://forums.guru3d.com/data/avatars/m/196/196284.jpg
There's really no reason for NVidia to get upset. (NVidia actually bought CUDA, they didn't create it themselves, btw. They also hired it's actual creator. Google did the same thing with Android. Bought the product and hired the creator.) CUDA relies on high-level languages that NVidia has no legal rights to. As long as AMD is "converting" source code, NVidia has no grounds to pursue them as NVidia has no rights to the source code. The only way NVidia would/should have any legal recourse, is if AMD is using the actual CUDA API itself.
#5190509
data/avatar/default/avatar33.webp
"The work AMD is doing to produce a high-performance compiler that sits below high-level programming models enables researchers to concentrate on solving problems and publishing groundbreaking research rather than worrying about hardware-specific optimizations." Does he really believe in what he say. He should say " Sorry scientists we failed to offer you proper tools to write your applications, we failed to offer you any technical support, we couldn't offer programming specialist to help you write coding lines and these greedy bastards at nVidia offer all that for free or for little money just to make you buy their hardware and because CUDA is open and expanding in every scientific and engineering field we decide to port CUDA to our hardware even if that mean you need to do some coding work and alot of debugging, So please buy some of our hardware, It's cheaper"
#5190521
https://forums.guru3d.com/data/avatars/m/258/258664.jpg
No, you need source to compile library. It is quite possible that this compiler will be free as most of AMD development tools. But one requires source PhysX CUDA code, that's in nVidia's hands. That 2 years old DirectCompute promise... I do not think nV will make PhysX available to run on AMD hardware even if it was effortless to them.
Oh they promised to put it on AMD hardware? Wasn't aware of that.
still use that crap? right now... no, as the green are much more efficient and AMD is still anecdotic in pro station... maybe in the future... ...wich is not so evident right now.
What I'm thinking too. If it lacks performance, it takes away only half the reason to buy a nvidia gpu for such purposes.
There's really no reason for NVidia to get upset. (NVidia actually bought CUDA, they didn't create it themselves, btw. They also hired it's actual creator. Google did the same thing with Android. Bought the product and hired the creator.) CUDA relies on high-level languages that NVidia has no legal rights to. As long as AMD is "converting" source code, NVidia has no grounds to pursue them as NVidia has no rights to the source code. The only way NVidia would/should have any legal recourse, is if AMD is using the actual CUDA API itself.
Normally CUDA libraries are property of nvidia, aren't they? Like trademarked programs, apis, code... so I'm not sure it's easy for other companies to recompile it, no matter if AMD makes the tool to do so available. Isn't it more like 'you can use the code but not alter it'?
#5190554
https://forums.guru3d.com/data/avatars/m/252/252478.jpg
Isn't it more like 'you can use the code but not alter it'?
well technically it's exactly that since a compiler would simply generate fresh binaries compatible with AMD hardware based on existing CUDA code i hope this works out for amd but i dont see how or why anyone would prefer a ported version of software that's bound to have its own set of issues that take forever to debug to native code. the price difference between the two vendors isnt that *big*
#5190574
https://forums.guru3d.com/data/avatars/m/258/258664.jpg
well technically it's exactly that since a compiler would simply generate fresh binaries compatible with AMD hardware based on existing CUDA code i hope this works out for amd but i dont see how or why anyone would prefer a ported version of software that's bound to have its own set of issues that take forever to debug to native code. the price difference between the two vendors isnt that *big*
Well then it should be possible, yet I wouldn't bet money on it. I wonder how nvidia will react to that, if they do at all.
#5190606
data/avatar/default/avatar26.webp
An other articles ... http://www.extremetech.com/extreme/218092-amds-boltzmann-initiative-targets-the-high-performance-computing-market
Just to be clear, Hipify Tools doesn’t run CUDA applications on AMD chips. Instead, it performs a source-to-source translation that’s meant to make it easy for developers to target either architecture. We asked AMD what the typical performance hit looked like for performing this task, and the company told us that in general-use cases, the performance hit is effectively zero. If a developer has specifically targeted a specific NV architecture with a great deal of optimization for each GPU, then it would take more time to optimize the same cases for GCN — but that the code would work out of the box, even in those cases.
Unlike what’s been reported by some publications, AMD does not execute CUDA on GCN, CUDA applications are not analyzed or reverse-engineered, and AMD is not compiling these applications into OpenCL. The point of HIP is to allow for a vendor-neutral approach that targets either NVCC or HCC.
http://www.extremetech.com/wp-content/uploads/2015/11/AMDHIP2b_575px-640x709.png
#5190834
https://forums.guru3d.com/data/avatars/m/196/196284.jpg
Normally CUDA libraries are property of nvidia, aren't they? Like trademarked programs, apis, code... so I'm not sure it's easy for other companies to recompile it, no matter if AMD makes the tool to do so available. Isn't it more like 'you can use the code but not alter it'?
Nvidia owns the libraries, yes, but the remaining source code for CUDA based applications still belongs to the creator of the applications.

Next Page