TensorRT with Python 3.8

AI & Data Science Deep Learning (Training & Inference) TensorRT
#1

Description

In my application I would like to use the newly introduced features of Python 3.8 together with TensorRT. Unfortunately, TensorRT for Python 3.8 is not available and information on how to solve this problem can not be found.

Environment

CUDA Version: 10.2
** System **: Jetson Xavier AGX
** Jetpack version **: 4.4.1

2 Likes
#2

Hi @patrick.nagelm8t1h,
Currently, the latest container release supports python 3.6


Please stay tuned to the forum and TRT site for the updates.

Thanks!

#3

I mentioned it myself, but this is not a solution. The same obsolete answer was given at the beginning of 2020 already. Can anyone say if this topic will be addressed soon or what the difficulties are there?

1 Like
#4

You should also still be able to install and run TensorRT under different environments, including Python 3.8. For me I installed from wheel for other python version and manually moved the packages to python3.8 local packages. But it is not guaranteed it will work flawlessly in all cases

1 Like
#5

@shunfan2 thanks for you response. Unfortunately this does not work for me:

ImportError: Python version mismatch: module was compiled for Python 3.6, but the interpreter version is incompatible: 3.8.6 (default, Nov 18 2020, 15:00:51)
1 Like
#7

Hi @patrick.nagelm8t1h,
I am afraid, this is the best possible information i could provide.
Kindly check the NVIDIA Page for updates.

Thanks!

#8

Hi

same problem with Python 3.9

greetings