Great article on CUDA. I had been looking for a cheap way to build a Massively parallel computer and I found it. Also plan to start applying CUDA to our modelling process at work.
One concern I have been trying to install the cuda toolkit on Ubuntu 9.10 and Fedora 12 and running into some road blocks. Specifically
make[1]: Entering directory `/home/wallacem/NVIDIA_GPU_Computing_SDK/C/src/convolutionTexture'
/usr/local/cuda/bin/../include/math_functions.h:422: error: inline function ‘int __signbit(double)’ cannot be declared weak
It continues but all the errors are similar has to do with glibc-headers package. The toolkit install provides some useful demo programs to check that cuda is working and a good starting point for development. I was wondering if you had any suggestions for getting around this problem. From my searching NVIDIA does not support newer version of operating systems quickly or this package is not compatible with CUDA?
