NPP missing with CUDA 3.2

I am trying to install OpenCV2.2 on a macbook pro (10.6.7). I have the CUDA driver (3.2), the CUDA toolkit, SDK for GPU computing and have verified the samples in /Developer/GPU Computing/C/darwin/release .
These run fine. However, when I run cmake with any option (Xcode,UNIX), the option to configure with CUDA always fails with the error “CUDA_NPP_LIBRARY_ROOT_DIR-NOTFOUND”.
I looked online to see if NPP is to be downloaded separately. I don’t see any options to download. I cannot use CUDA4.0 simply because openCV doesn’t support it.
I have been trying to install OpenCV unsuccessfully for nearly 3 days. Could somebody please help me out with this? I would greatly appreciate any pointers.

The CUDA_NPP library isn’t found as part of the CMake distribution and is somehow configured by the local OpenCV source. You should contact the OpenCV developers, probably from their sourceforge project that I found when googling OpenCV.

I will get in touch with OpenCV forum. But could you clarify something? Isn’t NPP the native library provided by nVidia? NPP download is separate for windows and linux. Am I missing something here?

Regards

I will get in touch with OpenCV forum. But could you clarify something? Isn’t NPP the native library provided by nVidia? NPP download is separate for windows and linux. Am I missing something here?

Regards