Which version of LInux is easiest to use?

[b]So I’ve tried CUDA on windows, epic fail because the instructions are so poor from NVidia and the tutorials provided over the net are for different OS so can’t apply.

Have tried recently on Ubuntu 10.10 64 bit using the following tutorial, a very detailed one.[/b]

[i]Howto: Install Nvidia CUDA in Ubuntu 10.04

However it keeps giving me errors,

make[1]: Entering directory /home/mohammed/NVIDIA_GPU_Computing_SDK/C/common' make[1]: Leaving directory /home/mohammed/NVIDIA_GPU_Computing_SDK/C/common’
make[1]: Entering directory /home/mohammed/NVIDIA_GPU_Computing_SDK/C/common' make[1]: Leaving directory /home/mohammed/NVIDIA_GPU_Computing_SDK/C/common’
make[1]: Entering directory /home/mohammed/NVIDIA_GPU_Computing_SDK/C/common' make[1]: Leaving directory /home/mohammed/NVIDIA_GPU_Computing_SDK/C/common’
make[1]: Entering directory /home/mohammed/NVIDIA_GPU_Computing_SDK/shared' make[1]: Leaving directory /home/mohammed/NVIDIA_GPU_Computing_SDK/shared’
make -C src/MonteCarloMultiGPU/
make[1]: Entering directory /home/mohammed/NVIDIA_GPU_Computing_SDK/C/src/MonteCarloMultiGPU' make[1]: Leaving directory /home/mohammed/NVIDIA_GPU_Computing_SDK/C/src/MonteCarloMultiGPU’
make -C src/quasirandomGenerator/
make[1]: Entering directory /home/mohammed/NVIDIA_GPU_Computing_SDK/C/src/quasirandomGenerator' make[1]: Leaving directory /home/mohammed/NVIDIA_GPU_Computing_SDK/C/src/quasirandomGenerator’
make -C src/deviceQueryDrv/
make[1]: Entering directory /home/mohammed/NVIDIA_GPU_Computing_SDK/C/src/deviceQueryDrv' /usr/bin/ld: cannot find -lcuda collect2: ld returned 1 exit status make[1]: *** [../../bin/linux/release/deviceQueryDrv] Error 1 make[1]: Leaving directory /home/mohammed/NVIDIA_GPU_Computing_SDK/C/src/deviceQueryDrv’
make: *** [src/deviceQueryDrv/Makefile.ph_build] Error 2

It’s got to the point where I’m almost about to give up on CUDA, all I want to be able to do is compile the samples, especially the template and bandwidthTest ones, at least then I can go on with some basic program.

I wanted to know for what is the best/easiest method to get the CUDA samples compiling on my system. I’ve given up on windows, so linux is my only option. I have an empty partition of around 50Gb if needed. ANy ideas?

(Note: this answer may need to be adjusted for your specific installation)

You need to set your LD_LIBRARY_PATH to include the cuda library directories.

in bash, Something like “export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/usr/local/cuda/lib” should do the trick.

If you don’t want to bother with that, you can stuff into your .login / .profile (.cshrc).

or

Put your cuda libraries into your system’s default path by adding them to ld.so.conf or adding a cuda.conf file to /etc/ld.so.conf.d.

I did the latter on my ubuntu system and after running ldconfig I was able to move on to the next challenge. :-)