ceccocats/tkDNN

Make error: Building on Jetson Nano

Closed this issue · 1 comments

casper-hansen commented

After doing cmake .. with success and doing make, I get the following error.

I have the following installed on the Jetson Nano with JetPack R32.5.1

  • CUDA 10.2
  • cuDNN 8.0
  • TensorRT 7.1.3
  • CMake 3.20
  • OpenCV 4.1.1-2
  • yaml-cpp 0.5.5

I have also tried running in my own Docker container but with the same result.

Do you have a guide on how to make it run on the Jetson Nano (or Jetson in general)? It works fine on my Ubuntu 18.04, but for some reason, I cannot get it running on the Jetson. Am I supposed to use the Docker container provided in the repository?

Thank you in advance!

mgk@mgk-desktop: tkDNN-yolo/build$ cmake ..
-- Found CUDNN: /usr/lib/aarch64-linux-gnu/libcudnn.so
-- Found CUDNN include: /usr/lib/aarch64-linux-gnu
-- Found NVINFER: /usr/lib/aarch64-linux-gnu/libnvinfer.so
-- Found NVINFER include: /usr/include/aarch64-linux-gnu
install dir:/usr/local
-- Configuring done
-- Generating done
-- Build files have been written to: /home/mgk/tkDNN-yolo/build
mgk@mgk-desktop: tkDNN-yolo/build$ make
[  1%] Building NVCC (Device) object CMakeFiles/kernels.dir/src/kernels/kernels_generated_upsample.cu.o
/home/mgk/tkDNN-yolo/src/kernels/upsample.cu:1:10: fatal error: kernels.h: No such file or directory
 #include "kernels.h"
          ^~~~~~~~~~~
compilation terminated.
CMake Error at kernels_generated_upsample.cu.o.cmake:220 (message):
  Error generating
  /home/mgk/tkDNN-yolo/build/CMakeFiles/kernels.dir/src/kernels/./kernels_generated_upsample.cu.o


CMakeFiles/kernels.dir/build.make:194: recipe for target 'CMakeFiles/kernels.dir/src/kernels/kernels_generated_upsample.cu.o' failed
make[2]: *** [CMakeFiles/kernels.dir/src/kernels/kernels_generated_upsample.cu.o] Error 1
CMakeFiles/Makefile2:583: recipe for target 'CMakeFiles/kernels.dir/all' failed
make[1]: *** [CMakeFiles/kernels.dir/all] Error 2
Makefile:135: recipe for target 'all' failed
make: *** [all] Error 2
casper-hansen commented

There was some issue on how I cloned the repository that apparently did not work. Thank you guys for your incredible repo!

I literally just copy-pasted the guide from the README and then it worked.

  • ❤️1