NVIDIA
Failure of TensorRT 10.5 to install when building on AGX Jetson
Description
Tried to install tensorrt on AGX Jetson
Environment
Python: 3.8 Torch: 2.0 (for Jetson) Jetpack 5.1.2 Pip: 24
Steps To Reproduce
pip install tensorrt -->
Commands or scripts:
pip install tensorrt==10.1 Looking in indexes: https://pypi.org/simple, https://pypi.ngc.nvidia.com Collecting tensorrt==10.1 Downloading tensorrt-10.1.0.tar.gz (16 kB) Preparing metadata (setup.py) ... done Collecting tensorrt-cu12 (from tensorrt==10.1) Downloading tensorrt-cu12-10.5.0.tar.gz (18 kB) Preparing metadata (setup.py) ... error error: subprocess-exited-with-error
× python setup.py egg_info did not run successfully.
│ exit code: 1
╰─> [6 lines of output]
Traceback (most recent call last):
File "
note: This error originates from a subprocess, and is likely not a problem with pip. error: metadata-generation-failed
× Encountered error while generating package metadata. ╰─> See above for output.
note: This is an issue with the package mentioned above, not pip. hint: See above for details.
NOTE Basically I cannot install tensorrt on AGX Jetson anymore, and I cannot find a solution for the given error message online.
I used 10.1 in this example but I also used the default 10.5.*
pip install tensorrt Looking in indexes: https://pypi.org/simple, https://pypi.ngc.nvidia.com Collecting tensorrt Downloading tensorrt-10.5.0.tar.gz (16 kB) Preparing metadata (setup.py) ... done Collecting tensorrt-cu12==10.5.0 (from tensorrt) Downloading tensorrt-cu12-10.5.0.tar.gz (18 kB) Preparing metadata (setup.py) ... error error: subprocess-exited-with-error
× python setup.py egg_info did not run successfully.
│ exit code: 1
╰─> [6 lines of output]
Traceback (most recent call last):
File "
note: This error originates from a subprocess, and is likely not a problem with pip. error: metadata-generation-failed
× Encountered error while generating package metadata. ╰─> See above for output.
note: This is an issue with the package mentioned above, not pip. hint: See above for details. (perivenv) orbital@orb-uig146:~/yolo$
raise RuntimeError("TensorRT does not currently build wheels for Tegra systems")
It is said that the right Jetson SDK installation is required.
I am encountering the same exact issue running Jetpack 6.1 (CUDA 12.6) on a NVIDIA Jetson Orin Nano Development Kit.
$ pip install tensorrt
Defaulting to user installation because normal site-packages is not writeable
Collecting tensorrt
Using cached tensorrt-10.5.0.tar.gz (16 kB)
Preparing metadata (setup.py) ... done
Collecting tensorrt-cu12==10.5.0 (from tensorrt)
Using cached tensorrt-cu12-10.5.0.tar.gz (18 kB)
Preparing metadata (setup.py) ... error
error: subprocess-exited-with-error
× python setup.py egg_info did not run successfully.
│ exit code: 1
╰─> [6 lines of output]
Traceback (most recent call last):
File "<string>", line 2, in <module>
File "<pip-setuptools-caller>", line 34, in <module>
File "/tmp/pip-install-s97on5w_/tensorrt-cu12_a329c527bf13469ea2e5b4cbec08f586/setup.py", line 69, in <module>
raise RuntimeError("TensorRT does not currently build wheels for Tegra systems")
RuntimeError: TensorRT does not currently build wheels for Tegra systems
[end of output]
note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed
× Encountered error while generating package metadata.
╰─> See above for output.
note: This is an issue with the package mentioned above, not pip.
hint: See above for details.
This totally precludes the ability to use torch2trt (as well as any other Python scripts that leverage this library) since that is dependent on being able to have a Python-accessible installation of TensorRT.
Never mind, I used the TensorRT .deb file for Jetson/Jetpack systems (instead of the TAR I was originally trying to make work) and that cleared the error I was originally having (where the system could not find TensorRT at all in the Python setup.py script that required it to be import'd) that resulted in the above errorcode. Hope this helps others on the similar setup I'm on- unsure if it works for OP's setup.
where did you get the tensorrt .deb file?
my nvidia sdk manager install a tar.gz file right now i get the same error.
k@ubuntu:~/Desktop/ven1$ pyenv virtualenv --system-site-packages ven2py3-11-11
k@ubuntu:~/Desktop/ven1$ pyenv virtualenvs
3.11.11/envs/ven2py3-11-11 (created from /home/k/.pyenv/versions/3.11.11)
3.12.8/envs/ven1py3-12-8 (created from /home/k/.pyenv/versions/3.12.8)
ven1py3-12-8 (created from /home/k/.pyenv/versions/3.12.8)
ven2py3-11-11 (created from /home/k/.pyenv/versions/3.11.11)
k@ubuntu:~/Desktop/ven1$ pyenv activate ven2py3-11-11
(ven2py3-11-11) k@ubuntu:~/Desktop/ven1$ pip --version
pip 24.0 from /home/k/.pyenv/versions/3.11.11/envs/ven2py3-11-11/lib/python3.11/site-packages/pip (python 3.11)
(ven2py3-11-11) k@ubuntu:~/Desktop/ven1$ pip install --upgrade pip
Requirement already satisfied: pip in /home/k/.pyenv/versions/3.11.11/envs/ven2py3-11-11/lib/python3.11/site-packages (24.0)
Collecting pip
Downloading pip-24.3.1-py3-none-any.whl.metadata (3.7 kB)
Downloading pip-24.3.1-py3-none-any.whl (1.8 MB)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.8/1.8 MB 20.6 MB/s eta 0:00:00
Installing collected packages: pip
Attempting uninstall: pip
Found existing installation: pip 24.0
Uninstalling pip-24.0:
Successfully uninstalled pip-24.0
Successfully installed pip-24.3.1
(ven2py3-11-11) k@ubuntu:~/Desktop/ven1$ pip check
No broken requirements found.
(ven2py3-11-11) k@ubuntu:~/Desktop/ven1$ tenssorrt --version
bash: tenssorrt: command not found
(ven2py3-11-11) k@ubuntu:~/Desktop/ven1$ tensorrt --version
bash: tensorrt: command not found
(ven2py3-11-11) k@ubuntu:~/Desktop/ven1$ python3 -m pip install --upgrade tensorrt
Collecting tensorrt
Using cached tensorrt-10.7.0.tar.gz (16 kB)
Installing build dependencies ... done
Getting requirements to build wheel ... done
Preparing metadata (pyproject.toml) ... done
Collecting tensorrt_cu12==10.7.0 (from tensorrt)
Using cached tensorrt_cu12-10.7.0.tar.gz (18 kB)
Installing build dependencies ... done
Getting requirements to build wheel ... error
error: subprocess-exited-with-error
× Getting requirements to build wheel did not run successfully.
│ exit code: 1
╰─> [20 lines of output]
Traceback (most recent call last):
File "/home/k/.pyenv/versions/ven2py3-11-11/lib/python3.11/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 353, in <module>
main()
File "/home/k/.pyenv/versions/ven2py3-11-11/lib/python3.11/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 335, in main
json_out['return_val'] = hook(**hook_input['kwargs'])
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/k/.pyenv/versions/ven2py3-11-11/lib/python3.11/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 118, in get_requires_for_build_wheel
return hook(config_settings)
^^^^^^^^^^^^^^^^^^^^^
File "/tmp/pip-build-env-bm3zizm1/overlay/lib/python3.11/site-packages/setuptools/build_meta.py", line 334, in get_requires_for_build_wheel
return self._get_build_requires(config_settings, requirements=[])
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/tmp/pip-build-env-bm3zizm1/overlay/lib/python3.11/site-packages/setuptools/build_meta.py", line 304, in _get_build_requires
self.run_setup()
File "/tmp/pip-build-env-bm3zizm1/overlay/lib/python3.11/site-packages/setuptools/build_meta.py", line 522, in run_setup
super().run_setup(setup_script=setup_script)
File "/tmp/pip-build-env-bm3zizm1/overlay/lib/python3.11/site-packages/setuptools/build_meta.py", line 320, in run_setup
exec(code, locals())
File "<string>", line 71, in <module>
RuntimeError: TensorRT does not currently build wheels for Tegra systems
[end of output]
note: This error originates from a subprocess, and is likely not a problem with pip.
error: subprocess-exited-with-error
× Getting requirements to build wheel did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.
(ven2py3-11-11) k@ubuntu:~/Desktop/ven1$ dpkg -l | grep nvinfer
ii libnvinfer-bin 10.3.0.30-1+cuda12.5 arm64 TensorRT binaries
ii libnvinfer-dev 10.3.0.30-1+cuda12.5 arm64 TensorRT development libraries
ii libnvinfer-dispatch-dev 10.3.0.30-1+cuda12.5 arm64 TensorRT development dispatch runtime libraries
ii libnvinfer-dispatch10 10.3.0.30-1+cuda12.5 arm64 TensorRT dispatch runtime library
ii libnvinfer-headers-dev 10.3.0.30-1+cuda12.5 arm64 TensorRT development headers
ii libnvinfer-headers-plugin-dev 10.3.0.30-1+cuda12.5 arm64 TensorRT plugin headers
ii libnvinfer-lean-dev 10.3.0.30-1+cuda12.5 arm64 TensorRT lean runtime libraries
ii libnvinfer-lean10 10.3.0.30-1+cuda12.5 arm64 TensorRT lean runtime library
ii libnvinfer-plugin-dev 10.3.0.30-1+cuda12.5 arm64 TensorRT plugin libraries
ii libnvinfer-plugin10 10.3.0.30-1+cuda12.5 arm64 TensorRT plugin libraries
ii libnvinfer-samples 10.3.0.30-1+cuda12.5 all TensorRT samples
ii libnvinfer-vc-plugin-dev 10.3.0.30-1+cuda12.5 arm64 TensorRT vc-plugin library
ii libnvinfer-vc-plugin10 10.3.0.30-1+cuda12.5 arm64 TensorRT vc-plugin library
ii libnvinfer10 10.3.0.30-1+cuda12.5 arm64 TensorRT runtime libraries
ii python3-libnvinfer 10.3.0.30-1+cuda12.5 arm64 Python 3 bindings for TensorRT standard runtime
ii python3-libnvinfer-dev 10.3.0.30-1+cuda12.5 arm64 Python 3 development package for TensorRT standard runtime
ii python3-libnvinfer-dispatch 10.3.0.30-1+cuda12.5 arm64 Python 3 bindings for TensorRT dispatch runtime
ii python3-libnvinfer-lean 10.3.0.30-1+cuda12.5 arm64 Python 3 bindings for TensorRT lean runtime
(ven2py3-11-11) k@ubuntu:~/Desktop/ven1$ pip install tensorrt
Collecting tensorrt
Using cached tensorrt-10.7.0.tar.gz (16 kB)
Installing build dependencies ... done
Getting requirements to build wheel ... done
Preparing metadata (pyproject.toml) ... done
Collecting tensorrt_cu12==10.7.0 (from tensorrt)
Using cached tensorrt_cu12-10.7.0.tar.gz (18 kB)
Installing build dependencies ... done
Getting requirements to build wheel ... error
error: subprocess-exited-with-error
× Getting requirements to build wheel did not run successfully.
│ exit code: 1
╰─> [20 lines of output]
Traceback (most recent call last):
File "/home/k/.pyenv/versions/3.11.11/envs/ven2py3-11-11/lib/python3.11/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 353, in <module>
main()
File "/home/k/.pyenv/versions/3.11.11/envs/ven2py3-11-11/lib/python3.11/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 335, in main
json_out['return_val'] = hook(**hook_input['kwargs'])
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/k/.pyenv/versions/3.11.11/envs/ven2py3-11-11/lib/python3.11/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 118, in get_requires_for_build_wheel
return hook(config_settings)
^^^^^^^^^^^^^^^^^^^^^
File "/tmp/pip-build-env-bjwyajgi/overlay/lib/python3.11/site-packages/setuptools/build_meta.py", line 334, in get_requires_for_build_wheel
return self._get_build_requires(config_settings, requirements=[])
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/tmp/pip-build-env-bjwyajgi/overlay/lib/python3.11/site-packages/setuptools/build_meta.py", line 304, in _get_build_requires
self.run_setup()
File "/tmp/pip-build-env-bjwyajgi/overlay/lib/python3.11/site-packages/setuptools/build_meta.py", line 522, in run_setup
super().run_setup(setup_script=setup_script)
File "/tmp/pip-build-env-bjwyajgi/overlay/lib/python3.11/site-packages/setuptools/build_meta.py", line 320, in run_setup
exec(code, locals())
File "<string>", line 71, in <module>
RuntimeError: TensorRT does not currently build wheels for Tegra systems
[end of output]
note: This error originates from a subprocess, and is likely not a problem with pip.
error: subprocess-exited-with-error
× Getting requirements to build wheel did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.
@kairin NOTE HERE: TensorRT does not currently build wheels for Tegra systems
@kairin NOTE HERE: TensorRT does not currently build wheels for Tegra systems
yes, aware of that... hence subsequently resorting to docker instead since there are prebuilt docker files.
jetson-containers although upon checking it is only 10.4.0
root@kkkORIN:/opt/comf# dpkg -l | grep tensorrt
root@kkkORIN:/opt/comf# cd ..
root@kkkORIN:/opt# ls
comf nvidia protobuf-3.20.3-cp310-cp310-linux_aarch64.whl
ComfyUI oneTBB tinycudann
diffusers opencv torch2trt
diffusers-0.33.0.dev0-py3-none-any.whl opencv_contrib vhacdx
manifold opencv_contrib_python-4.10.0.86-cp310-cp310-linux_aarch64.whl xformers
numpy-2.2.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl polyscope xformers-0.0.30+56be3b5.d20250122-cp310-cp310-linux_aarch64.whl
root@kkkORIN:/opt# pip install tensorrt
Looking in indexes: https://pypi.jetson-ai-lab.dev/jp6/cu126, https://pypi.ngc.nvidia.com
Requirement already satisfied: tensorrt in /usr/local/lib/python3.10/dist-packages (10.4.0)
root@kkkORIN:/opt#
Hi, just wanted to share what I have learned after 2 days of searching and trying to install tensorrt on my jetson orin machine that I work on:
setup:ubuntu linux on aarch64 cpu with tegra chip.
cuda12, jetpack 6 and TensorRT are allready installed on the system.
And I do not have sudo permissions.
YES these information are important for installation (that does not work).
so if you want to install any package then I recommend that you trust your pip since it does every thing it can to install the desired package. With that said now let me tell you why I could not install tensorrt on my system.
with the following command:
pip install --index-url https://pypi.nvidia.com tensorrt
pip searches the nvidia pypi index to find any compatible wheels or buildable .tar.gz files, so it finds tensorrt==10.8.0.43 which is not a wheel that can easily be installed but a .tar.gz file that needs to be built locally to generate a wheel compatible with the system then install the wheel. BUT this package cannot be built because I am on a.. yes you guessed it a tegra system.
So I searched the nvidia pypi index manually and opened every setup.py in every .tar.gz file to find any version which does not assert that the system is not tegra and does build on tegra systems. And I found out that the tensorrt 8.6.1 can be built on tegra systems but when trying to install it a get this error message
ERROR: Could not find a version that satisfies the requirement tensorrt_libs==8.6.1 (from versions: none)
so I looked at the pypi nvidia index and found only wheels (no .tar.gz) for tensorrt-libs==8.6.1 that are not compatible with aarch64 soo it cannot be installed using pip. RACAP: no tensorrt-libs==8.6.1 compatible with aarch64 > no tensorrt==8.6.1 (version can be installed on tegra systems) CONCLUSION: USE DOCKER!!
luckily for me I found out there is allready a version of tensorrt that is 8.6.2 installed as system-site-package on my machine, and I just needed to create a virtual environment with system-site-packages enabled. somehow someone was able to install it on the machine but it is not any more in pypi.nvidia.com index.
[EDIT]
found this one tensorrt for jetson hardware , I needed to install it using apt instead of pip => but I never considered such threads because I do not have sudo permissions. BUT it was already installed on my system and this explains why it worked when I enable system-site-packages as it also worked for the person in the above thread.
