Revision history [back]

Hi, I'm on ubuntu 13.04, have successfully compiled opencv 2.4.6.1 with the latest cuda 5.5. At least I thought it was a success. Because when I try to compile the examples in /opencv-2.4.6.1/samples/c, all I get is:

/usr/bin/ld: warning: libnppc.so.5.5, needed by /usr/local/cuda/lib64/libnppi.so.5.5, not found (try using -rpath or -rpath-link)

And all the consequences of the above:

/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `nppGetMaxThreadsPerSM'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `nppGetStream'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `Stream'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `nppGetGpuComputeCapability'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `Event'
/usr/local/lib/libopencv_gpu.so: undefined reference to `nppSetStream'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `nppGetGpuNumSMs'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `nppGetPointerMemoryType'
collect2: error: ld returned 1 exit status

This being my first use of cuda and compilation with opencv, I don't know where to go from this. The only libnppc.so.5.5 I have found was for linux suse. Is there any thing I didn't set up properly when installing cuda (did with the .deb packages from nvidia site)? Cuda examples compile properly, opencv compiled and installed without any error.

Ludo

Hi, I'm on ubuntu 13.04, have successfully compiled opencv 2.4.6.1 with the latest cuda 5.5. At least I thought it was a success. Because when I try to compile the examples in /opencv-2.4.6.1/samples/c, all I get is:

/usr/bin/ld: warning: libnppc.so.5.5, needed by /usr/local/cuda/lib64/libnppi.so.5.5, not found (try using -rpath or -rpath-link)

And all the consequences of the above:

/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `nppGetMaxThreadsPerSM'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `nppGetStream'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `Stream'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `nppGetGpuComputeCapability'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `Event'
/usr/local/lib/libopencv_gpu.so: undefined reference to `nppSetStream'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `nppGetGpuNumSMs'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `nppGetPointerMemoryType'
collect2: error: ld returned 1 exit status

This being my first use of cuda and compilation with opencv, I don't know where to go from this. The only libnppc.so.5.5 I have found was for linux suse. Is there any thing I didn't set up properly when installing cuda (did with the .deb packages from nvidia site)? Cuda examples compile properly, opencv compiled and installed without any error.

LudoEdit: I've just tried to compile an example (houghlines.cpp) in the gpu section of the samples (opencv-2.4.6.1/samples/), with:

g++ houghlines.cpp -o houghlines `pkg-config --cflags --libs opencv` -lopencv_gpu

and it compiled and I could run the program all right. The above problem with libnppc remains with cpu examples (such as those in samples/c and samples/cpp) and any other older program I wrote...

Hi, I'm on ubuntu 13.04, have successfully compiled opencv 2.4.6.1 with the latest cuda 5.5. At least I thought it was a success. Because when I try to compile the examples in /opencv-2.4.6.1/samples/c, all I get is:

/usr/bin/ld: warning: libnppc.so.5.5, needed by /usr/local/cuda/lib64/libnppi.so.5.5, not found (try using -rpath or -rpath-link)

And all the consequences of the above:

/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `nppGetMaxThreadsPerSM'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `nppGetStream'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `Stream'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `nppGetGpuComputeCapability'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `Event'
/usr/local/lib/libopencv_gpu.so: undefined reference to `nppSetStream'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `nppGetGpuNumSMs'
/usr/local/cuda/lib64/libnppi.so.5.5: undefined reference to `nppGetPointerMemoryType'
collect2: error: ld returned 1 exit status

This being my first use of cuda and compilation with opencv, I don't know where to go from this. The only libnppc.so.5.5 I have found was for linux suse. Is there any thing I didn't set up properly when installing cuda (did with the .deb packages from nvidia site)? Cuda examples compile properly, opencv compiled and installed without any error.

Edit: I've just tried to compile an example (houghlines.cpp) in the gpu section of the samples (opencv-2.4.6.1/samples/), with:

g++ houghlines.cpp -o houghlines `pkg-config --cflags --libs opencv` -lopencv_gpu

and it compiled and I could run the program all right. The above problem with libnppc remains with cpu examples (such as those in samples/c and samples/cpp) and any other older program I wrote...