There is a patent from NVIDIA for utilizing GPU. If I want to use CUDA Toolkit in my program, which NVIDIA is giving for free, am I infringing the patent? I am not directly using the described method, but I think that is happening inside of the whole program that is actually NVIDIA providing (and I even don't have access to that). Even with accepting the CUDA EULA Agreement (which permits "Install and use the SDK") does it still mean that I still can be sued from NVIDIA for using the patent? Or NVIDIA is fine by this because what's the point providing something for free to all and in the same time having ammunition to sue someone?
This really has to do with the license of the CUDA Toolkit.– Eric S ♦Feb 26, 2022 at 13:54
Hi. Well, they are not saying anything explicit regarding patents except that they disclaim any warranty (which can be read here). They, however, say that it is permitted to install and use the SDK. I don't know does that mean we are fine regarding the patent. I can't use the patent from my side directly because I don't have an access to that. The problem is what if a function that user is calling uses the patent internally.– MarkoFeb 26, 2022 at 16:18
Your use of the Toolkit is governed by the license. Higher up in the page you linked describes the license. I think it is pretty clear you can use the software without worrying about patents, but I am not a lawyer. Licensing questions are off topic here buy you could try the Law SE site.– Eric S ♦Feb 26, 2022 at 22:33
NVIDIA's business model is selling hardware GPUs. Reading the EULA section on licensing would seem to indicate that you can freely use the CUDA Toolkit so long as you abide by the license restrictions which include not reverse engineering the code or using the software with non-NVIDIA hardware.
As such I believe this is really about licensing and not patents. I am not a lawyer and this is not a legal opinion. You might get some value posting the question on the SE Law site. Also, simply contacting NVIDIA and asking may prove helpful.