Running centos 7 w gtx 980m device w nvidia 355.11 driver
Installed cuda toolkit 7.5
Whe i try to build oceanfft cuda-opengl-cufft sample getting link error ld cannot find libGL.so
Non-GL samples build and run fine
Also curiously the glxinfo tool runs fine
and if I do ldd /usr/bin/glxinfo it shows
libGL.so.1
Tried installing freeglut3, latest glew tarball (get same error when I try to build that),
Also Mesa GL and Glu development libraries, but still not fixed.
Want to start over and make sure Im getting all required packages/dependencies in correct order.
What is the right way? Cant find complete instructions in nvidia toolkitThx.
Well this post fixed my error: the libGL.so symlink was dangling, repointed to libGL.so.1 which points to libGL.so.355.11
https://devtalk.nvidia.com/default/topic/928174/opengl/-usr-bin-ld-cannot-find-lgl-collect2-error-ld-returned-1-exit-status/
However, pretty sure I installed a bunch of stuff I don’t really need while trying to fix this.
So I’d still appreciate a clear explaniation of what the recommended minimum development setup configuration is to support CUDA with OpenGL integration.
(i.e. is installing freeglut3 and/or glew-1-13.0 redundant? what about the mesa GL devel libraries?)
New to Linux and CUDA/OpenGL development, so everything is confusing. Thanks.