TensorRT Installation on AWS EC2 Ubuntu 18.04 Error

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

Description

A clear and concise description of the bug or issue.

Hello,

I am trying to install TensorRT on ec2 Deep Learning AMI (Ubuntu 18.04) Version 48.0.

Since CUDA and cuDNN are installed, I only download the TensorRT installation file which is ‘nv-tensorrt-repo-ubuntu1804-cuda10.2-trt7.0.0.11-ga-20191216_1-1_amd64.deb’.

But, when I run the installation file, I got following error.

I tried the install these libraries (libnvinfer7) but couldnt manage it.

What should I do?

Or could someone suggest me a better way to install TensorRT to ec2 ubuntu system?

Thanks

Image from iOS
Image from iOS1920×1440 272 KB

Environment

TensorRT Version: 7.0
GPU Type:
Nvidia Driver Version:
CUDA Version: 10.2
CUDNN Version:
Operating System + Version:
Python Version (if applicable):
TensorFlow Version (if applicable): 1.15
PyTorch Version (if applicable):
Baremetal or Container (if container which image + tag):

Relevant Files

Please attach or include links to any models, data, files, or scripts necessary to reproduce your issue. (Github repo, Google Drive, Dropbox, etc.)

Steps To Reproduce

Please include:

  • Exact steps/commands to build your repro
  • Exact steps/commands to run your repro
  • Full traceback of errors encountered
#2

Hi,
Please refer to the installation steps from the below link if in case you are missing on anything
https://docs.nvidia.com/deeplearning/tensorrt/install-guide/index.html
Also, we suggest you to use TRT NGC containers to avoid any system dependency related issues.
https://ngc.nvidia.com/catalog/containers/nvidia:tensorrt

Thanks!

#5

Hello,

@NVES thanks for your reply,

I started a new Deep Learning AMI (Ubuntu 18.04) Version 48.0 instance and try to use TRT NGC but I got an error.

Firstly, I installed the ngc and match my API key. Then, I pull the TensorRT container (docker pull nvcr.io/nvidia/tensorrt:21.08-py3).

After that, I try to run this container by nvidia-docker run -it --rm -v local_dir:nvcr.io/nvidia/tensorrt:21.08-py3 command, I got an error like this.

tensorrt error
tensorrt error1920×493 68.2 KB

Why I am getting this error?

What should I do different or what did I do wrong way?

Thanks

#6

Hi,

Looks like you mentioned volume path wrong as local_dir:container_dir. Please mention local dir path as local path on your aws machine. Also please give container directory in correct format, ex: /home/data/

Thank you.

#7

Please refer to the installation steps from the below link if in case you are missing on anything

I followed the instructions in that doc and arrived at the same error that skilic shows in the first post: a bunch of lines saying that tensorrt depends on various libnv* libraries but that those won’t be installed. There is no reason given for why. Is there a separate package that needs to be installed before tensorrt that is not mentioned in the instructions?

EDIT: I figured out the issue on my end, the instruction say to make sure you have a valid version of CUDA, which I had (driver CUDA 11.0) but for some reason the toolkit version didn’t match the driver version (nvcc --version returned 10.0). Updating the toolkit to 11.0 and then redoing the steps succeeded.

1 Like