Problem compiling SDK

Hello,
I doesn’t succeed to compile the SDK after installing CUDA. I hope anybody could help me to fix it, looking on internet doesn’t help me so much…
I can compile and execute cuda code using nvcc in command line with my own code, that’s working well.
But when I try to compile with make in NVIDIA_GPU_Computing_SDK/, I have the following error :

make[1]: entering the directory « /home/alexandre/NVIDIA_GPU_Computing_SDK/shared »
make[1]: leaving the directory « /home/alexandre/NVIDIA_GPU_Computing_SDK/shared »
make[1]: entering the directory « /home/alexandre/NVIDIA_GPU_Computing_SDK/C »
make[2]: entering the directory « /home/alexandre/NVIDIA_GPU_Computing_SDK/C/common »
make[2]: leaving the directory « /home/alexandre/NVIDIA_GPU_Computing_SDK/C/common »
make[2]: entering the directory « /home/alexandre/NVIDIA_GPU_Computing_SDK/C/common »
make[2]: leaving the directory « /home/alexandre/NVIDIA_GPU_Computing_SDK/C/common »
make[2]: entering the directory « /home/alexandre/NVIDIA_GPU_Computing_SDK/C/common »
make[2]: leaving the directory « /home/alexandre/NVIDIA_GPU_Computing_SDK/C/common »
make[2]: entering the directory « /home/alexandre/NVIDIA_GPU_Computing_SDK/shared »
make[2]: leaving the directory « /home/alexandre/NVIDIA_GPU_Computing_SDK/shared »
make[2]: entering the directory « /home/alexandre/NVIDIA_GPU_Computing_SDK/C/src/volumeRender »
make[2]: /opt/bin/nvcc : command not found
make[2]: *** [obj/x86_64/release/volumeRender_kernel.cu.o] Error 127
make[2]: leaving the directory « /home/alexandre/NVIDIA_GPU_Computing_SDK/C/src/volumeRender »
make[1]: *** [src/volumeRender/Makefile.ph_build] Erreur 2
make[1]: leaving the directory « /home/alexandre/NVIDIA_GPU_Computing_SDK/C »
make: *** [all] Error 2

I found that in fact, nvcc is in “/opt/cuda/bin/”, so I added “/cuda” everywhere it was missing in .bash_profile. It was still not working. I took a look to the makefile containing the definition of the compiler nvcc (file “C/common/common.mk” calling apparently by all makefiles) and “/cuda” was missing in the address given for nvcc. But adding this, I have a permission denied and the examples from SDK still do not work.

I work it on a DELL T1700 with linux.

I don’t understand why my code is running well whereas SDK doesn’t recognize nvcc. Do you know what I can do to fix that problem ?

Hi,
Which distro and cuda verison did you use?

Hi, I use the version 5.0.35-4ubuntu1 of cuda on ubuntu 13.04.