undefined symbol: __cudaPopCallConfiguration

[euler16]

I understand that the repository was targeted towards CUDA 8. I have CUDA 10 on my system and after making relevant changes in the Makefile , I am able to build this package. But when I run test_shiftnet.py I am getting the above mentioned error.

[euler16]

The output after running make is

Separate cpp shared library that will be loaded to the extern C ffi
nvcc -arch=sm_30 -gencode=arch=compute_30,code=sm_30 -gencode=arch=compute_50,code=sm_50 -gencode=arch=compute_52,code=sm_52 -gencode=arch=compute_60,code=sm_60 -gencode=arch=compute_61,code=sm_61 -gencode=arch=compute_61,code=compute_61 -c -o build/shiftnet_cuda_kernels.so src/shiftnet_cuda_kernels.cu -x cu -Xcompiler -fPIC -shared -Iinclude
python build_ffi.py
generating /tmp/tmpNa46bR/_shiftnet_cuda.c
setting the current directory to '/tmp/tmpNa46bR'
running build_ext
building '_shiftnet_cuda' extension
creating mnt
creating mnt/data
creating mnt/data/rajivratn
creating mnt/data/rajivratn/nil_astu
creating mnt/data/rajivratn/nil_astu/end2end
creating mnt/data/rajivratn/nil_astu/end2end/videonly
creating mnt/data/rajivratn/nil_astu/end2end/videonly/shiftnet_cuda_v2
creating mnt/data/rajivratn/nil_astu/end2end/videonly/shiftnet_cuda_v2/src
gcc -pthread -B /home/rajivratn/anaconda2/compiler_compat -Wl,--sysroot=/ -fno-strict-aliasing -g -O2 -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -DWITH_CUDA -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/TH -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/THC -I/usr/local/cuda/include -I/mnt/data/rajivratn/nil_astu/end2end/videonly/shiftnet_cuda_v2/include -I/home/rajivratn/anaconda2/include/python2.7 -c _shiftnet_cuda.c -o ./_shiftnet_cuda.o
gcc -pthread -B /home/rajivratn/anaconda2/compiler_compat -Wl,--sysroot=/ -fno-strict-aliasing -g -O2 -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -DWITH_CUDA -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/TH -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/THC -I/usr/local/cuda/include -I/mnt/data/rajivratn/nil_astu/end2end/videonly/shiftnet_cuda_v2/include -I/home/rajivratn/anaconda2/include/python2.7 -c /mnt/data/rajivratn/nil_astu/end2end/videonly/shiftnet_cuda_v2/src/shiftnet_cuda.c -o ./mnt/data/rajivratn/nil_astu/end2end/videonly/shiftnet_cuda_v2/src/shiftnet_cuda.o
gcc -pthread -shared -B /home/rajivratn/anaconda2/compiler_compat -L/home/rajivratn/anaconda2/lib -Wl,-rpath=/home/rajivratn/anaconda2/lib -Wl,--no-as-needed -Wl,--sysroot=/ ./_shiftnet_cuda.o ./mnt/data/rajivratn/nil_astu/end2end/videonly/shiftnet_cuda_v2/src/shiftnet_cuda.o /mnt/data/rajivratn/nil_astu/end2end/videonly/shiftnet_cuda_v2/build/shiftnet_cuda_kernels.so -L/home/rajivratn/anaconda2/lib -lpython2.7 -o ./_shiftnet_cuda.so

[HaFred]

Hi @euler16 , would you kindly opensource the makefile for CUDA10? Running into same problems here. Thanks a lot.

[euler16]

Hi @HaFred I couldn't make this work for CUDA10.

[xin-h963]

Hi @euler16 , I have a similar problem. undefined symbol: __cudaRegisterFatBinaryEnd
Did you solve it? Thank you.