Failed loading CUDA library when trying to use ffmpeg with nvenc

Question

On Ubuntu 14.04 I am following the instruction guide from NVIDIA's site to get ffmpeg running using GPU encoding.

I followed the instructions (as far as I can tell), installed the NVIDIA SDK local .dep, installed the cuda package and the nvidia-352 package, then compiled ffmpeg.

However, when I try to do the following call:

ffmpeg -i ~/test/StreamInstructions.mp4 -vcodec nvenc -b:v 5M -f avi output.mp4

ffmpeg tells me Failed loading CUDA library.

Searching google shows me the source code for this error message originates on the line of code: dl_fn->cuda_lib = dlopen("libcuda.so", RTLD_LAZY);.

This says to me (who hasn't done much c++ or linux in the past decade) that it can't find the libcuda.so shared library.

Doing a search of the file system I see that libcuda.so is in the following locations:

/usr/local/cuda-7.5/doc/man/man7/libcuda.so.7
/usr/local/cuda-7.5/targets/x86_64-linux/lib/stubs/libcuda.so
/usr/lib/i386-linux-gnu/libcuda.so.1
/usr/lib/i386-linux-gnu/libcuda.so
/usr/lib/i386-linux-gnu/libcuda.so.352.63
/usr/lib/x86_64-linux-gnu/libcuda.so.1
/usr/lib/x86_64-linux-gnu/libcuda.so.352.63

I can't find any other results in my searching to help resolve this issue, and because of that I feel like I might be doing something wrong on the linux level rather than the ffmpeg level.

Anyone have any suggestions on where to go from here?

Edit: I just noticed that ldconfig -p actually shows that it can find the libcuda.so library, yet I'm still getting errors that it can't find the library:

 ldconfig -p | grep cuda
        libicudata.so.52 (libc6,x86-64) => /usr/lib/x86_64-linux-gnu/libicudata.so.52
        libcuda.so.1 (libc6,x86-64) => /usr/lib/x86_64-linux-gnu/libcuda.so.1
        libcuda.so.1 (libc6) => /usr/lib/i386-linux-gnu/libcuda.so.1
        libcuda.so (libc6) => /usr/lib/i386-linux-gnu/libcuda.so

Answer 1

NOTE:
Back up your old LD_LIBRARY_PATH first!
This command will place the value of the environment variable LD_LIBRARY_PATH into your home directory in the file "LD_LIBRARY_PATH.old"
#>$ echo $LD_LIBRARY_PATH >> ~/LD_LIBRARY_PATH.old

MY FIX:
I had this issue when I installed the NVidia/Cuda drivers in Fedora. I had to update my library path:
#>$ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/path/to/cuda/lib

REVERT CHANGES:
#>$ export LD_LIBRARY_PATH=$(cat ~/LD_LIBRARY_PATH.old)