tkDNN
tkDNN copied to clipboard
ceccocats
'error adding symbols' during "make" in jetson nano
when I make tkDNN in jetson nano, I met unexpected error message as below:
kibum@kibum-desktop:~/tkDNN/build$ make CMake Warning (dev) at CMakeLists.txt:21: Syntax Warning in cmake code at column 30
Argument not separated from preceding token by whitespace. This warning is for project developers. Use -Wno-dev to suppress it.
-- Found CUDNN: /usr/local/cuda/lib64/libcudnn.so -- Found CUDNN include: /usr/local/cuda/include -- Found NVINFER: /usr/lib/aarch64-linux-gnu/libnvinfer.so -- Found NVINFER include: /usr/include/aarch64-linux-gnu Eigen DIR: /usr/include/eigen3 install dir:/usr/local -- Configuring done -- Generating done -- Build files have been written to: /home/kibum/tkDNN/build [ 1%] Linking CXX shared library libkernels.so [ 13%] Built target kernels [ 14%] Linking CXX shared library libtkDNN.so /usr/local/cuda/lib64/libcudnn.so: error adding symbols: File in wrong format collect2: error: ld returned 1 exit status CMakeFiles/tkDNN.dir/build.make:667: recipe for target 'libtkDNN.so' failed make[2]: *** [libtkDNN.so] Error 1 CMakeFiles/Makefile2:750: recipe for target 'CMakeFiles/tkDNN.dir/all' failed make[1]: *** [CMakeFiles/tkDNN.dir/all] Error 2 Makefile:148: recipe for target 'all' failed make: *** [all] Error 2
Please let me know how to solve this error....
Jetpack: 4.5 Cuda: 10.2 cudnn: 8.0.0 opencv: 4.1.1 TensorRT: 7.1.3.0
are you cross compiling ?if so then you can try this guide https://forums.developer.nvidia.com/t/solved-libcudart-so-error-adding-symbols-file-in-wrong-format/71084/3
Also, which version of cmake are you using? tkDNN requires cmake>= 3.15 and the default in the Jetpack is way smaller.
@mive93 I already check my cmake version is 3.18.2 which was upgraded before. so, I think it is not the reason to make this problem.
@perseusdg Thank you for your information. I checked the cross compiling issue in my jetson nano. but it is not easy to find which folders are conflicted since the issues are discussed with old version in 2years ago. Anyway, I'll try again :-)