Questions on installation (Driver and SDK compile

I am installing CUDA on Fedora 13.

I have installed the nVidia video card driver from nVidia website. But I see on the CUDA development page, there is another driver called development driver. Do you need to install that also? The video card driver version is 260.19.29, while the dev driver version is 260.19.21. Is it OK to install a old dev driver on a new video card driver?

Another problem I am having is, when I am using make to complie examples in SDK, the linker reports error:

/usr/bin/ld: cannot find -lGLU

collect2: ld returned 1 exit status

make[1]: *** [../../bin/linux/release/marchingCubes] Error 1

make[1]: Leaving directory `/home/szli/NVIDIA_GPU_Computing_SDK/C/src/marchingCubes'

make: *** [src/marchingCubes/Makefile.ph_build] Error 2

But I see the library file libGLU.so.1 in /usr/lib. When I copy the file and name it libGLU.so, the linker is able to link to this library but cannot find some other libraries. Is there anyway to let the linker to find the correct .so.1 (or .so.6, etc) correctly? Thanks.

Did you see the Known Issues (Section III.) in the CUDA_SDK_Release_Notes.txt file? They address this specific problem, but the suggested solution is just to create a symbolic link to the file that you do have (with the .1 extension). There is no mention of a general solution of the kind you’re asking about, so I expect the answer is just to do this for every file you encounter that causes this problem. Hopefully that won’t be too many files …