Closed SeokhyunKim closed 7 years ago
Try changing that -l/usr/local/cuda/lib to -L/usr/local/cuda/lib and check that /usr/local/cuda/lib actually contains the cud library files (those that are required by the -lcuda parameter)
Yes it solved by your comment. Thank you, mbertini. It would be good if the configure script in ccv/lib is modified for this problem, I think.
@SeokhyunKim @mbertini can i run ccv on CPU and can you provide me a step by step procedure for installation of ccv in ubuntu
Hello. I'm sw engineer of Software R&D Center, Samsung electronics.
Recently, I tested libccv on NVIDIA Jetson TK1 board. I wanted to compile cnnclassify.c in ccv/bin folder, and got the following error messages:
make -C ../lib make[1]: Entering directory
/home/ubuntu/workspace/ccv/lib' make[1]: Nothing to be done for
all'. make[1]: Leaving directory `/home/ubuntu/workspace/ccv/lib' gcc -o bbffmt bbffmt.o -L"../lib" -lccv -L/usr/local/lib -lm -lpng -ljpeg -lcuda -lcudart -lcublas -l/usr/local/cuda/lib /usr/bin/ld: cannot find -lcudart /usr/bin/ld: cannot find -lcublas /usr/bin/ld: cannot find -l/usr/local/cuda/lib collect2: error: ld returned 1 exit status make: *\ [bbffmt] Error 1This may be simple problem, and if you want I could make a simple bug fix patch. Thank you.