I’ve read a few topics around the question of CUDA library redistribution -some as far back as 2015. Nothing really addresses whether there’s a viable method to ship application complete code built against CUDA & C++ libraries without having end-users download the complete developers toolkit? There is talk of CUDA statically linking the necessaries dll’s but nothing confirming if that happened or not? There is also talk of reading the EULA and the redistribution dll’s but nothing that describes where these files are supposed to be installed to and whether you need to add Environment variable to Windows. It would be really good to have something definitive here -even some examples of what needs to happen to pull this off. At the moment I am compelled to ask my end users to download the complete CUDA developer toolkit which seems overkill. Appreciate any help/pointers on this. Thanks!