compiling multiple .cu files into one .cubin file?

I am trying to compile two .cu files (using nvcc --cubin) and have the output be a single .cubin file. A colleague of mine is able to do this using v. 0.8 version of CUDA on a Windows machine, but we haven’t been able to determine which compiler flag allows this. I am running v. 1.0 on a Linux machine. (I have also tried it with v. 1.1 on a Linux machine.)

Can anyone tell me if this is even an option anymore? (And if so, how does one go about it?)


You can simply #include “” into your other .cu file to make them look as one bigger .cu to nvcc.

Thank you!