Hey Joe Hoeller, thanks for the compliment and glad that the post has been helpful to you so far! I haven’t run into this issue myself, but I’ll try to be as helpful as I can. Seems like you’re up against a dependency issue related to CUDA’s use of the OpenGL library, not a symlink problem. It’s weird that you’re running in to this on CentOS, because all the Googling that I did related to people having this issue with Ubuntu.

Quick question: does nvcc -V give you the expected output? Want to make sure that we’re good up to that point in the guide.

I checked out this NVIDIA forums post: https://devtalk.nvidia.com/default/topic/993862/cuda-8-0-missing-required-libraries-on-ubuntu-16-04/. It seems like you have a couple options:

1. Try running make -k, which will bypass sample builds that depend on OpenGL
2. Install the necessary dependencies. With Ubuntu, apparently, that looks like this: sudo apt-get install freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libgl1-mesa-glx libglu1-mesa libglu1-mesa-dev libglfw3-dev libgles2-mesa-dev. That user than ran make with GLPATH=/usr/lib make. Obvious apt-get is the Ubuntu package manager, so you’d need to look up how to install these with yum install.

Hope that gives you a couple pointers! I’m curious to know what the issue is and how you resolved it. Once we debug this I’m happy to update the post for other users who might run into the same thing :)