I admit being new to GPU/CUDA programming, but even so developing a trivial “hello world” dll that exports functions that use CUDA, e.g. return a string with the current device identification, seems unreasonably complex, to say the least. Makes one wonder if it is possible at all.
Please notice I am not worried about the actual CUDA functions to get that info. The SDK includes lots of examples that do that. My concern is how one builds A REGULAR DLL that exports some function, say “myCudaHelloWorld()” which function uses the GPU and does whatever useful. I am a bit surprised there are no clear, step by step examples in the SDK. Also most thread advice I read here doesnt work, maybe because the authors assume you have knowledge that a newbie of course does not have.
So, is there any such simple-dll-step-by-step example somewhere?
Anyone from NVidia that reads these forums? <img src=‘http://hqnveipbwb20/public/style_emoticons/<#EMO_DIR#>/crying.gif’ class=‘bbc_emoticon’ alt=’:’(’ />