cuda 9.1 Makefile for cudaDecodeGL in Samples fails

Entering directory ‘/home/frank/NVIDIA_CUDA-9.1_Samples/3_Imaging/cudaDecodeGL’
“/usr/local/cuda-9.1”/bin/nvcc -ccbin g++ -m64 -Xcompiler -no-pie -gencode arch=compute_30,code=compute_30 -o cudaDecodeGL FrameQueue.o ImageGL.o VideoDecoder.o VideoParser.o VideoSource.o cudaModuleMgr.o cudaProcessFrame.o dynlink_cuda.o dynlink_nvcuvid.o videoDecodeGL.o -L/usr/lib/nvidia-375 -lGL -lGLU -lX11 -lglut -lcuda -L…/…/common/lib/linux/x86_64 -lcudart -lnvcuvid -lGLEW
/usr/bin/ld: cannot find -lnvcuvid
collect2: error: ld returned 1 exit status
Makefile:404: recipe for target ‘cudaDecodeGL’ failed
make[1]: *** [cudaDecodeGL] Error 1
make[1]: Leaving directory ‘/home/frank/NVIDIA_CUDA-9.1_Samples/3_Imaging/cudaDecodeGL’
Makefile:52: recipe for target ‘3_Imaging/cudaDecodeGL/Makefile.ph_build’ failed
make: *** [3_Imaging/cudaDecodeGL/Makefile.ph_build] Error 2

The “cannot find -lnvcuvid” error is discussed in a number of postings on this forum if you care to look for them.


I will hunt them down.

I have retreated to trying to build everything from scratch to try and get Digits to run, so every single problem gets fixed.

here’s one example thread:

in a nutshell, libnvcuvid is installed by the driver, and the exact location for this may vary. If you locate that file ( then you can usually “fix” the makefile for this sample code, by modifying the -L switch.

if you don’t actually care about this sample code, you can just issue

make -k

instead of


and the samples should all build, except for those that have a problem.

not sure why any of this would have anything to do with DIGITS

Thank you so much

I did LIBRARY_PATH=/usr/lib/nvidia-384:$LIBRARY_PATH make
and the problem went away. marvellous thanks fir the pointer to the solutions