# Trying to install pytorch with cuda in jetson orin NX **URL:** \ **Category:** Jetson Orin NX\ **Tags:** pytorch\ **Created:** [July 30, 2024, 5:02pm UTC](https://forums.developer.nvidia.com/t/trying-to-install-pytorch-with-cuda-in-jetson-orin-nx/301726 "2024-07-30T17:02:33Z")\ **Posts on this page:** 6\ **Page:** 1
**Author:** ![juliocesargcosta123](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/juliocesargcosta123/32/251286_2.png) [@juliocesargcosta123](https://forums.developer.nvidia.com/u/juliocesargcosta123)\ **Post date:** [July 30, 2024, 5:02pm UTC](https://forums.developer.nvidia.com/t/trying-to-install-pytorch-with-cuda-in-jetson-orin-nx/301726/1 "2024-07-30T17:02:33Z")
So, im trying to install the pytorch with cuda support in my jetson orin NX (dev kit), i already have done a fresh install using the SDK Manager and i have jetpack 6.0, and ubuntu 22.04, i tried to follow the instructions from the nvidia manual, and configured a virtual enviroment: [https://docs.nvidia.com/deeplearning/frameworks/install-pytorch-jetson-platform/index.html](https://docs.nvidia.com/deeplearning/frameworks/install-pytorch-jetson-platform/index.html), the issue is that im getting an error in the pre-requisites installation of the libopenblas-dev like the following: sudo apt-get -y update; sudo apt-get -y install python3-pip libopenblas-dev; … OTHER LINKS … Ign:9 [https://packagecloud.io/headmelted/codebuilds/debian](https://packagecloud.io/headmelted/codebuilds/debian) stretch InRelease Err:10 [https://packagecloud.io/headmelted/codebuilds/debian](https://packagecloud.io/headmelted/codebuilds/debian) stretch Release 404 Not Found [IP: 2600:1f1c:2e5:6901:e375:73d1:461b:b889 443] Reading package lists… Done E: The repository ‘[https://packagecloud.io/headmelted/codebuilds/debian](https://packagecloud.io/headmelted/codebuilds/debian) stretch Release’ does not have a Release file. N: Updating from such a repository can’t be done securely, and is therefore disabled by default. N: See apt-secure(8) manpage for repository creation and user configuration details. Reading package lists… Done Building dependency tree… Done Reading state information… Done libopenblas-dev is already the newest version (0.3.20+ds-1). python3-pip is already the newest version (22.0.2+dfsg-1ubuntu0.4). 0 upgraded, 0 newly installed, 0 to remove and 18 not upgraded., Besides that i kept following the instructions and successfully installed the torch in the right version as it shows in the pip list, but when i try to import torch in a python program i get the following error: python3 Python 3.10.12 (main, Mar 22 2024, 16:50:05) [GCC 11.4.0] on linux Type “help”, “copyright”, “credits” or “license” for more information. > > > import torch > > > Traceback (most recent call last): > > > File “/home/cedri/Documents/torch\_test/venv/lib/python3.10/site-packages/torch/ **init**.py”, line 225, in \_load\_global\_deps > > > ctypes.CDLL(global\_deps\_lib\_path, mode=ctypes.RTLD\_GLOBAL) > > > File “/usr/lib/python3.10/ctypes/ **init**.py”, line 374, in **init** > > > self.\_handle = \_dlopen(self.\_name, mode) > > > OSError: libnvToolsExt.so.1: cannot open shared object file: No such file or directory During handling of the above exception, another exception occurred: Traceback (most recent call last): File “”, line 1, in File “/home/cedri/Documents/torch\_test/venv/lib/python3.10/site-packages/torch/ **init**.py”, line 288, in \_load\_global\_deps() File “/home/cedri/Documents/torch\_test/venv/lib/python3.10/site-packages/torch/ **init**.py”, line 246, in \_load\_global\_deps \_preload\_cuda\_deps(lib\_folder, lib\_name) File “/home/cedri/Documents/torch\_test/venv/lib/python3.10/site-packages/torch/ **init**.py”, line 168, in \_preload\_cuda\_deps raise ValueError(f"{lib\_name} not found in the system path {sys.path}") ValueError: libcublas.so.\*[0-9] not found in the system path [‘’, ‘/usr/lib/python310.zip’, ‘/usr/lib/python3.10’, ‘/usr/lib/python3.10/lib-dynload’, ‘/home/cedri/Documents/torch\_test/venv/lib/python3.10/site-packages’], I suspect that could be a problem related to the pre-requistes installation error, but as the link gives a 404 error, i dont have a clue on how to solve this. ---
**Author:** ![AastaLLL](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/aastalll/32/14043_2.png) [@AastaLLL](https://forums.developer.nvidia.com/u/AastaLLL)\ **Post date:** [July 31, 2024, 3:41am UTC](https://forums.developer.nvidia.com/t/trying-to-install-pytorch-with-cuda-in-jetson-orin-nx/301726/3 "2024-07-31T03:41:40Z")
Hi, Have you edited the source list? The below errors should occur when running the `sudo apt-get -y update`. ```auto Ign:9 https://packagecloud.io/headmelted/codebuilds/debian stretch InRelease Err:10 https://packagecloud.io/headmelted/codebuilds/debian stretch Release ``` But it’s not a default path in our JetPack 6.0. Do you add them manually? Could you edit the `/etc/apt/sources.list` and remove the link to see if it helps? Thanks. ---
**Author:** ![juliocesargcosta123](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/juliocesargcosta123/32/251286_2.png) [@juliocesargcosta123](https://forums.developer.nvidia.com/u/juliocesargcosta123)\ **Post date:** [August 2, 2024, 8:31pm UTC](https://forums.developer.nvidia.com/t/trying-to-install-pytorch-with-cuda-in-jetson-orin-nx/301726/4 "2024-08-02T20:31:52Z")
Hi AastaLLL, Thank you in advance, I discovered that the dep links were from another file within the /etc/apt folder, and most likely because of the VS Code installation I did previously, so when i removed these links the error no longer occurred. Going further into the torch installation manual again, another error occurs when I try to import the torch library: > from torch.\_C import \* # noqa: F403 > ImportError: libcupti.so.12: cannot open shared object file: No such file or directory I suspect that this error ocurred because of the cuda version that i had installed in the jetson which is currently the cuda-12.2. Also looking to the $LD\_LIBRARY\_PATH which points to the cuda installation folder i found a lot of libcu\*.so.12 files but not the libcupti.so.12 file or similar. Another thing that i would like to ask is about the step showing below from the tutorial, i could not find the llvm-8 directory in usr/lib/. > $ export LD\_LIBRARY\_PATH=/usr/lib/llvm-8/lib:$LD\_LIBRARY\_PATH Best Regards, Julio César ---
**Author:** ![AastaLLL](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/aastalll/32/14043_2.png) [@AastaLLL](https://forums.developer.nvidia.com/u/AastaLLL)\ **Post date:** [August 28, 2024, 3:18am UTC](https://forums.developer.nvidia.com/t/trying-to-install-pytorch-with-cuda-in-jetson-orin-nx/301726/7 "2024-08-28T03:18:03Z")
Hi, Sorry for the late update. Do you get PyTorch work? Thanks. ---
**Author:** ![juliocesargcosta123](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/juliocesargcosta123/32/251286_2.png) [@juliocesargcosta123](https://forums.developer.nvidia.com/u/juliocesargcosta123)\ **Post date:** [August 31, 2024, 3:01pm UTC](https://forums.developer.nvidia.com/t/trying-to-install-pytorch-with-cuda-in-jetson-orin-nx/301726/8 "2024-08-31T15:01:01Z")
Hi AastaLLL, Yes, i could make pytorch work just by changing the pytorch version in the virtual enviroment. To help others that can have the same problem as i, here is the specs that i used: Jetson Orin NX Jetpack: 6.0 [L4T 36.3](https://developer.nvidia.com/embedded/jetson-linux-r363) CUDA runtime version: 12.2 (Was installed via SDK Manager) Pytorch version: [torch-2.3.0a0+ebedce2.nv24.02-cp310-cp310-linux\_aarch64.whl](https://developer.download.nvidia.cn/compute/redist/jp/v60dp/pytorch/torch-2.3.0a0+ebedce2.nv24.02-cp310-cp310-linux_aarch64.whl) (in a python venv) To verify if the pytorch is working with CUDA integration, you can use a python file with following code: > import torch > print("CUDA is available: ", torch.cuda.is\_available()) Also verifying with the jtop (jtop installation guide [https://jetsonhacks.com/2023/02/07/jtop-the-ultimate-tool-for-monitoring-nvidia-jetson-devices/](https://jetsonhacks.com/2023/02/07/jtop-the-ultimate-tool-for-monitoring-nvidia-jetson-devices/)) command in the GPU tab if the gpu resources are being used. Best Regards, Júlio César ---
**Author:** ![system](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/system/32/68080_2.png) [@system](https://forums.developer.nvidia.com/u/system)\ **Post date:** [September 14, 2024, 3:01pm UTC](https://forums.developer.nvidia.com/t/trying-to-install-pytorch-with-cuda-in-jetson-orin-nx/301726/9 "2024-09-14T15:01:34Z")
Closed on September 14, 2024, 3:01pm UTC This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.