I can compile without problems my CUDA program but when I try to execute it I get this error:
./Reconnaissance_De_Media: error while loading shared libraries: libcufft.so.2: cannot open shared object file: No such file or directory
I am novice with GPGPU and CUDA programming and I don’t figure out why this error occurs.
Thanks for any help
I have been having the same problems. I compiled all the samples and tried to run convolutionFFT2D and i get this error:
libcufft.so.6.0: cannot open shared object file: No such file or directory
I’ve looked around everywhere, can’t find anything that works so I tried something myself to troubleshoot. I copied all the libcufft.so.x.y.z files from my usr/local/cuda-6.0/lib64 directory to the usr/lib/x86_64-linux-gnu/ directory and now I am getting this error:
./convolutionFFT2D: error while loading shared libraries: libcufft.so.6.0: wrong ELF class: ELFCLASS32
It seems as if the problem is in these files. Does anyone know what might be the solution?
give the path of respected libcufft.so.2 file… its linking problem… dont copy the file to ur directory… give the path where it originally occur …
But where do I change the path of the file? Is there a makefile in the samples which has the directory for the libcufft binaries which I can edit to fix the problem?
Generally speaking, you would want to add the path for any DSOs (“dynamic libraries”, “shared objsects”) required by an application to the environment variable LD_LIBRARY_PATH.