Hello everyone,
I hope you’re all doing well. I’m currently in the process of installing TensorRT and would greatly appreciate your insights on a matter.
My question is whether downgrading some of the packages might be necessary to ensure a proper installation of TensorRT. I’ve been facing a bit of uncertainty regarding the correctness of my TensorRT installation. While I haven’t encountered the TensorRT-related warning in TensorFlow, I did encounter a series of challenges to get TensorRT up and running.
Presently, the latest PyTorch version (2.0.1) only functions with CUDA 11.8. Consequently, I had to carry out downgrades for CUDA, cuDNN, and TensorRT. I must mention that these packages were added to the Ubuntu repository. However, I received prompts to upgrade them, each time I sudo apt update && sudo apt upgrade, which I temporarily mark them by placing them “on hold” to avoid any potential conflicts.
If any of you could share your expertise and advice in this matter, it would be immensely helpful. Your experience and insights would contribute greatly to resolving my situation.
When I type:
$ dpkg -l | grep nvinfer
I get:
ii libnvinfer-dispatch-dev 8.6.1.6-1+cuda12.0 amd64 TensorRT development dispatch runtime libraries
ii libnvinfer-dispatch8 8.6.1.6-1+cuda12.0 amd64 TensorRT dispatch runtime library
ii libnvinfer-headers-dev 8.6.1.6-1+cuda12.0 amd64 TensorRT development headers
ii libnvinfer-headers-plugin-dev 8.6.1.6-1+cuda12.0 amd64 TensorRT plugin headers
ii libnvinfer-lean-dev 8.6.1.6-1+cuda12.0 amd64 TensorRT lean runtime libraries
ii libnvinfer-lean8 8.6.1.6-1+cuda12.0 amd64 TensorRT lean runtime library
ii libnvinfer-plugin8 8.6.1.6-1+cuda11.8 amd64 TensorRT plugin libraries
ii libnvinfer-vc-plugin-dev 8.6.1.6-1+cuda12.0 amd64 TensorRT vc-plugin library
ii libnvinfer-vc-plugin8 8.6.1.6-1+cuda12.0 amd64 TensorRT vc-plugin library
ii libnvinfer8 8.6.1.6-1+cuda11.8 amd64 TensorRT runtime libraries
As you can see there are still some packages with cuda12.0 version! Should I downgrade each one of them as well?! How can I know I have a proper Installation of TensorRT anyway? I haven’t seen much of difference in performance, yet!
Environment
TensorRT Version: 8.6.1
GPU Type: 4060
Nvidia Driver Version:
CUDA Version: 535.86.05
CUDNN Version: 8.9.0
Operating System + Version: Ubuntu 22.04
Python Version (if applicable): 3.10
TensorFlow Version (if applicable): 2.13
PyTorch Version (if applicable): 2.0.1
Thank you in advance for your time and assistance.