Jetson orin AGX pytorch error "OSError: libcudnn.so.8: cannot open shared object file: No such file or directory"

Autonomous Machines Jetson & Embedded Systems Jetson AGX Orin
,
1

Hi, I am setting environment on new Jetson agx orin.

My CUDA version is 11.4, cuDNN is 8.1.1.

I installed pytorch by following “pytorch for jetson”

My pytorch version is 1.12.0a0+2c916ef.nv22.3.

However, if I import torch on python, I got error bellow

Python 3.8.10 (default, Nov 14 2022, 12:59:47) 
[GCC 9.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/rise/.local/lib/python3.8/site-packages/torch/__init__.py", line 198, in <module>
    _load_global_deps()
  File "/home/rise/.local/lib/python3.8/site-packages/torch/__init__.py", line 151, in _load_global_deps
    ctypes.CDLL(lib_path, mode=ctypes.RTLD_GLOBAL)
  File "/usr/lib/python3.8/ctypes/__init__.py", line 373, in __init__
    self._handle = _dlopen(self._name, mode)
OSError: libcudnn.so.8: cannot open shared object file: No such file or directory

I also set PATH and LD_LIBRARY_PATH like bellow.

PATH="${PATH}:${HOME}/.local/bin" 
export PATH=$PATH:/usr/local/cuda-11.4/bin
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-11.4/lib64:/usr/local/lib

I did everything I found on google to solve this error, but still I’m stuck on this issue.

How can I solve it?

3

Hi,

Please help to double-check the cuDNN version in your environment with the following command.

$ cat /usr/include/aarch64-linux-gnu/cudnn_version_v8.h

Thanks.

4

I got error like bellow.

cat: /usr/include/aarch64-linux-gnu/cudnn_version_v8.h: No such file or directory

Is cuDNN is not installed properly?

I installed with bellow commands.

tar –xzvf cudnn-11.2-linux-x64-v8.1.1.33.tgz
sudo cp cuda/include/cudnn*.h /usr/local/cuda/include
sudo cp -P cuda/lib64/libcudnn* /usr/local/cuda/lib64
sudo chmod a+r /usr/local/cuda/include/cudnn*.h /usr/local/cuda/lib64/libcudnn*

I can check my version by bellow command

command 1

cat /usr/local/cuda/include/cudnn_version.h | grep CUDNN_MAJOR -A 2

result

#define CUDNN_MAJOR 8
#define CUDNN_MINOR 1
#define CUDNN_PATCHLEVEL 1
--
#define CUDNN_VERSION (CUDNN_MAJOR * 1000 + CUDNN_MINOR * 100 + CUDNN_PATCHLEVEL)

#endif /* CUDNN_VERSION_H */

command2

ldconfig -N -v $(sed 's/:/ /' <<< $LD_LIBRARY_PATH) 2>/dev/null | grep libcudnn

result

	libcudnn_ops_train.so.8 -> libcudnn_ops_train.so.8.1.1
	libcudnn_adv_infer.so.8 -> libcudnn_adv_infer.so.8.1.1
	libcudnn.so.8 -> libducnn.so.8
	libcudnn_ops_infer.so.8 -> libcudnn_ops_infer.so.8.1.1
	libcudnn_cnn_train.so.8 -> libcudnn_cnn_train.so.8.1.1
	libcudnn_adv_train.so.8 -> libcudnn_adv_train.so.8.1.1
	libcudnn_cnn_infer.so.8 -> libcudnn_cnn_infer.so.8.1.1
5

Hi,

The steps look incorrect.

You will need a Jetson package for Orin but the package you used is for the x86 environment.
Could you reflash the system and install the components from JetPack and try it again.

Thanks

6

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.