TensorRT 5.1.2 installation on Jetson TX2 board

Hi

Can anyone tell me how to upgrade from TRT 5.0.6 to 5.1.2.
Currently I’m using Jetpack 4.2 which gives TRT 5.0.6, CUDA 10.0, cudnn 7.3.

Whenever I follow the steps in https://docs.nvidia.com/deeplearning/sdk/tensorrt-install-guide/index.html#installing-debian I get the following errors

The following packages have unmet dependencies.
tensorrt : Depends: libnvinfer5 (= 5.1.2-1+cuda10.0) but it is not going to be installed
Depends: libnvinfer-dev (= 5.1.2-1+cuda10.0) but it is not going to be installed
Depends: libnvinfer-samples (= 5.1.2-1+cuda10.0) but it is not going to be installed
E: Unable to correct problems, you have held broken packages.

I don’t think it’s because of the broken packages as I’ve downloaded the debian file from nvidia website and tried again.

Also which version of CUDA & cudnn should I use for TRT 5.1.2 ?

[quote]

Hi

Can anyone tell me how to upgrade from TRT 5.0.6 to 5.1.2.
Currently I’m using Jetpack 4.2 which gives TRT 5.0.6, CUDA 10.0, cudnn 7.3.

Whenever I follow the steps in https://docs.nvidia.com/deeplearning/sdk/tensorrt-install-guide/index.html#installing-debian I get the following errors

The following packages have unmet dependencies.
tensorrt : Depends: libnvinfer5 (= 5.1.2-1+cuda10.0) but it is not going to be installed
Depends: libnvinfer-dev (= 5.1.2-1+cuda10.0) but it is not going to be installed
Depends: libnvinfer-samples (= 5.1.2-1+cuda10.0) but it is not going to be installed
E: Unable to correct problems, you have held broken packages.

I don’t think it’s because of the broken packages as I’ve downloaded the debian file from nvidia website and tried again.

Also which version of CUDA & cudnn should I use for TRT 5.1.2 ?[/quote

Hi @AastaLLL

Do you guys have anything on this.

Hello,

Did your try to follow the instructions in [url]https://docs.nvidia.com/deeplearning/sdk/tensorrt-install-guide/index.html#upgrading-5-51x[/url]?

Also, here’s the support matrix:
[url]https://docs.nvidia.com/deeplearning/dgx/tf-trt-user-guide/index.html#matrix[/url]

Thanks.

Im afraid that tensorrt can only be installed through Jetpack which means you cant install yourself with .deb file.Because these files are compiled for x86-64 architecture not arm64 which TX2 used.

Based on this article:

https://docs.nvidia.com/deeplearning/sdk/tensorrt-install-guide/index.html#add-install

We need to use JetPack in order to install TensorRT and actually JetPack is bundled with certain version of TensorRT.