Member

These fixes should have really been made with the CUDA 8 branch but somehow got missed. So in reality, compute_6x was never getting compiled.

Ref #1485

Member Author

Forgot to add, the CUDA samples also do not get compiled for compute_6x.

This can be verified by

cd /usr/local/cuda-8.0/samples/0_Simple/vectorAdd
make

Member

@shehzan10 can you also enable OpenMP? I wasn't able to compile examples and tests even on belle (Debian 8) with the devel branch

Member

I think this was tested on an incorrect version of CUDA. I installed the CUDA 8 RC and pulled in your changes to test again. I'll merge if they pass.

Member Author

No I tested with CUDA 8 for certain. I even tested the cuda 8 examples. nvcc - - help shows no computer with 6x.

Member

I sent you a personal message on our chat about this.