Tensorflow 2.10 and libnvinfer.so.8

General Discussion
#1

Hi, I show you the computer info:

lspci | grep VGA
01:00.0 VGA compatible controller: NVIDIA Corporation TU116 [GeForce GTX 1650] (rev a1)
08:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Picasso/Raven 2 [Radeon Vega Series / Radeon Vega Mobile Series] (rev c9)


and the issue:

![Screenshot from 2022-10-17 18-11-17|690x166](upload://xn9WiyrTcOYoWY96NmdAi44QyDH.png)

I installed `tensorflow 2.10`, `vndia-smi` show I have installed `cuda 11.7`, but in the pic say `CUDA 11.1`, why? 

Also, it say that it cant not load:

`libnvinfer.so.7`, but `tensorflow 2.10` install `libnvinfer.so.8`

I already fix the  path

export LD_LIBRARY_PATH="$LD_LIBRARY_PATH:/usr/lib/cuda-11.7/targets/x86_64-linux/lib/libcudnn"

Can  Anybody  help me?

Thanks,
#2

Have you followed tested build configurations? Thank you

#4

I try install tensorflow 2.10, CUDA 11.2 y cuDNN 8.1, like the first line in the configuration, in an env conda, but dont install them.

Maybe I should test install CUDA 11.2 y cuDNN 8.1 out of the env, directly in the system.

#6

Maybe I should test install CUDA 11.2 y cuDNN 8.1 out of the env, directly in the system.

Please try and let us know? Thank you

#8

It is difficult to install tf2.10 with conda env

#10

@tianliang,

Have you tried using Miniconda?

Miniconda is the recommended approach for installing TensorFlow with GPU support. It creates a separate environment to avoid changing any installed software in your system. This is also the easiest way to install the required software especially for the GPU setup. Please follow step by step instructions. Thank you!

#12

I could install tensorflow=2.10 with conda force and it installed its CUDA and cuDNN for default. It late a lot but finally it installed. Now tf detect the gpu and all works. Thank you for your tips!