It's a little bit complicated to start GPGPU programming on Linux. Things come to be difficult even begin from installing a new graphics driver. And here is my personal experience on how to solve the problems.
To make it clear enough, I separated the configuration process as the following steps:
Install the driver for your Linux distribution.
Install CUDA SDK.
LDvariables in the environment should be updated.
To update LD variable, create file "/etc/ld.so.conf.d/cuda.conf" and add the following content:
/etc/local/cuda/lib64for 64-bit systems).Then apply it by running
To update PATH variable, add
:/usr/local/cuda/bininto the file