Closed zhuziyu-edward closed 3 years ago
Hey @zhuziyu-edward,
So as _tensorflowcatkin and _tensorflow_roscpp are already included in this project, this shouldn't be too hard problem.
First install the required CUDA components, namely Cuda, Cudnn and NCCL. For me a quite helpful guide was this. Then please check your CUDA and CUDNN versions with:
nvcc --version
and cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2
And replace your path variables and version numbers here in this Cmake file.
Then running the catkin build with catkin build tensorflow_ros_cpp
in your catkin workspace root should work without an error. Please provide me further information if you are stuck with any of the steps or getting an error.
Note: Cublas header are changed above Cuda10.0, so better choose a previous release, or you have to use some workaround for _tensorflow_roscpp
Thanks a lot, I have successfully installed and run your code.
Happy to hear that! I will close the issue now.
Hi, I am trying to run your program, but I find it hard to compile tensorflow_ros_cpp. Could you please provide more detailed installation guide for us to follow? I am using ubuntu 16.04, do I have to manually compile tensorflow for this tensorflow_ros_cpp?