Tensor RT installation on NVIDIA Drive PX2 OS

Autonomous Vehicles DRIVE PX2 DRIVE - Linux
, ,
#1

Dear Team,

I am trying to install TensorRT on Drive PX2 OS Tegra A, but failing to get the TensorRT build.
I tried to install x86_64 and amd64 architecture TensorRT which failed as board as arch64 architecture.

I need arch64-based TensorRT version 3.0.2 or 3.0.4 or the latest version of the Tensor RT arch64-based installation file and installation guide for directly installing tensorRT On target device.

I have a few more queries regarding the environment setup on the target directly, issues faced are below:

  1. failed to install anaconda
  2. failed to install python-based dependencies for model porting
  3. failed to setup tensorRT

Let me know if it is possible to set up the model requirements for python-based dependencies directly on board OS.

Kindly do the needful.

Thanks and Regards,
Vyom Mishra

#3

Hi @vyom.mishra

I need arch64-based TensorRT version 3.0.2 or 3.0.4 or the latest version of the Tensor RT arch64-based installation file and installation guide for directly installing tensorRT On target device.

You can not install TensorRT as a standalone package. Also, TensorRT does not have python bindings on DRIVE platform. You need to use C++ APIs.
You need to flash the target using sdkmanager to get last release(DRIVE OS 5.0.10.3 with DriveWorks SDK ).

May I know what is the current DRIVE version installed on target?

#4

This post was flagged by the community and is temporarily hidden.