How to tell nvcc to avoid conflict between /usr/include and gcc's /opt/local/include/gcc47/c++/ on O...

I believe that latest version of cuda I can use on geforce 750m is cuda3.0.
And I think that I need GCC from 2012 for this version of cuda, and that is GCC4.7

however, when i try it i get conflict (probably with existing clang). G++ on it’s own works and compiles programs without problems. However, in combination with nvcc, compilation fails:

mbp-2:cuda$ nvcc -ccbin=/opt/local/bin/g+±mp-4.7 hello_world.cu
/opt/local/include/gcc47/c++/cstdlib:107:9: error: ‘atexit’ is already declared in this scope

GeForce 750m can use CUDA 7.5

follow the instructions in the linux install guide:

[url]Installation Guide Linux :: CUDA Toolkit Documentation

Unfortunately title got truncated.
Operating system OS X, 10.11.6 (15G1004)

When I tried CUDA 7.5 before, i got:

“cuda driver version is insufficient for cuda runtime version”

after the successful installation.

Then you haven’t installed things correctly. It’s not an indication that CUDA 7.5 won’t work with that GPU.

Please re-read the mac installation guide:

[url]http://docs.nvidia.com/cuda/cuda-installation-guide-mac-os-x/index.html#abstract[/url]

Hi txbob, I followed the instructions, and I ended up with a new error, CUDA 7.5 and Xcode 7:
https://devtalk.nvidia.com/default/topic/967208/cuda-setup-and-installation/-os-x-cuda-driver-version-is-insufficient-for-cuda-runtime-version/